Changelog

Every feature, improvement, and fix we ship to Scholarly, newest first.

November 7, 2026

Build 3D models with the best model

Turn on "Use best model" when you create a 3D model to build it with our most capable model, for 2 AI Creation Credits.
In chat, ask for the best model and the confirm card shows the 2-credit cost before anything starts.
Standard builds stay at 1 credit. The best model is on paid plans.

Page through documents right in chat

Slides, study guides, mind maps, story books, and every other document chat makes now open inside the message, the way 3D models do.
Flip pages with the arrows or your keyboard, and press Full page to open the document itself.
On phones, tap the first page to open it.

Get help from any page, no account needed

A help button now sits in the corner of every Scholarly website and Help Center page.
Ask Scholarly AI anything and get an answer from our help guides, with links to the guides it used.
Send our team a message from the same window, with your AI chat attached, and we reply by email.
If an answer doesn't help, tap thumbs down to reach a person.

Smaller changes

On the Free plan, a first chat answer is shorter and ends with one question that takes the lesson further.
When a creation is over your plan's limit, chat now starts teaching the topic right away and asks how you want to continue.
Confirm cards in chat now highlight the credit cost so you see the price before you say yes.
Videos and podcasts made from chat now cost exactly what the create window shows for the same length and voice, and chat tells you the credits each one used.
Opening past conversations on a new chat no longer hides your pinned chats and "Pick up where you left off".
Recordings where you speak more than one language, or switch between them, are now transcribed instead of failing with "no speech detected".
When chat starts a video, podcast, deck, or other creation for you, it now suggests one useful next step from the same material while you wait.
Flashcard decks made in chat now open inside the message: flip each card and step through the deck without leaving the conversation.
When a Premium plan runs out of weekly AI Creation Credits, the limit notice now offers Laureate, with 80 credits a week, and opens the upgrade confirmation in one step.
The phone welcome screen is shorter, with one less line between you and your first upload.
Making a podcast on a phone now fits on one screen: pick your options in a short list and press Create podcast.
If you close the file picker without choosing a file, the upload window offers to paste a link, a YouTube URL, or text instead.
On the Free plan, the video lecture creator shows how many more files Premium lets you add to one lecture.
Study guides made from part of a document name the pages they haven't covered yet.
Slides and flashcard decks show what the full version adds, and a quiz at its limit shows the next question with its answers locked.
Some new subscribers can choose yearly Laureate billing in the plan picker, with the full annual price shown before checkout.
Some Free accounts see a shorter plan picker on desktop that names what they were trying to make and lists Premium in five plain lines.
A new account whose browser could not finish opening onboarding no longer gets stuck reloading the page.
Quizzes on a shared deck now stop at the Free quiz limit with the usual limit screen, instead of opening the plan picker after every answer.
Scholarly loads again in Safari versions older than 16.4.

November 6, 2026

Remake a video lecture in another style

A new "Make another version" bar under your finished lecture gathers Another style, New voice, and Another language in one place.
Another style remakes the lecture from the same sources and settings in the style you pick, with each style's preview and the credit cost shown before you confirm.
Your original lecture stays as it is.

"Use my script as written" reads every word

A video made from your own script now narrates it word for word, in order, including reference lists.
Your script sets the video's length, and you see that length and its credit cost before you create.
A script longer than 45 minutes ends at a passage break, and one press makes Part 2 from exactly where Part 1 stopped.
Works with PDFs, Word files, and several files at once, in Standard, Freestyle, Math, and Custom Design videos.

Smaller changes

Google Drive, Dropbox and Notion now share one Connections tab when you add sources, with a dot on each app you've already connected.
Shared links now ask readers who aren't signed in to create an account first, then bring them straight back to what was shared.
A podcast's Make next row now fits on one line: Video, Summary, Flashcards and More, with New voices beside them at the same height.
Deleting your account no longer reports a failure when the deletion actually went through.
A recently deleted recording no longer offers create actions that could not run. Restore it to make something from it.
Uploading an audio file to a meeting that's larger than your plan allows now tells you right away and shows the plan that takes it, instead of uploading three times and then failing.
When a creation can't use a source, the reason now shows in your language instead of a generic error in Italian, French, Spanish, German, Portuguese and other languages.
A podcast's end card now appears when the episode finishes, even if a question card is still open, and offers the longer episode when one is available.
Speaker notes from a PowerPoint with hidden slides now stay with the right slide in chat and in everything you make from it.
Videos, podcasts, slides, flashcards and exams now teach at the level your material is written for: a school chapter in words a child follows, a graduate text without re-teaching the basics, and the audience or tone you ask for wins.
After you upgrade, the Slides length slider now moves to the paid default instead of staying at the Free length.
Flashcard citations now open the PDF at the cited page, and videos (including YouTube) and recordings at the cited moment.
Videos, documents and research now start with what you asked Scholarly to remember, so your saved preferences apply every time without an extra lookup.
When you message a video while it's being made, it now replies and finishes instead of re-checking a video that was already done.
PDF documents such as reports, worksheets and cheat sheets finish sooner, with fewer rebuilds when a table or page needs a fix.
Long 1080p video lectures now retry on their own if their workspace ends mid-render, so they no longer come out shorter than the length you picked.
A story book's Download menu now includes Page images (every page as a picture, in one zip) and files to publish it on Amazon KDP: a Kindle eBook for any story book, plus a paperback interior and wraparound cover for a square one.
Fewer slides in AI Slides and video lectures end in cut-off text, and comparisons of three or more things are planned as tables instead of turning into a plain list.
When a long video lecture's faster render crashes, it now tries again with its narration kept instead of being cut shorter.

November 5, 2026

Study materials that follow your request more closely

Tables in documents and cheat sheets now print right under their heading instead of drifting to a later page.
Videos, guides, and worksheets teach the subject directly instead of commenting on your file.
Exams mix true and false answers instead of leaning on "true".
5-minute podcasts now run closer to the length you choose.
Slide decks end on a recap of the key takeaways.

See and sign out your devices

Settings > Account now lists every device signed in to your account, with its approximate location and when it was last active.
Sign out any single device, or every device except the one you're using.
We email you when your account signs in on a new device. If it wasn't you, one press signs out everywhere and helps you set a new password.

Mark up any image

Draw, highlight, add arrows, boxes, numbered steps, and text on an image before you send it in chat.
Mark up images in a conversation, including ones chat made or found, and add the marked-up copy to your next message.
Blur private details, crop, and erase marks, with undo and redo. It works with touch on your phone too.
In Image Studio, draw on the parts you want changed and the edit follows your marks.

Smaller changes

The image markup controls scroll on phones so Undo and Redo stay within reach.
Canceling image markup returns you to the image you were viewing.
Quiz titles are written in your quiz's language.
Research reports keep every item you paste into your request, and saved tables and charts keep their file type.
Story books tell the whole story, ending included, within 12 pages.
Worksheet answer keys start on their own page.
Math animation videos show chapter summaries in the lecture's language, and long equations stay on screen.
Diagrams no longer repeat their subtitle as a caption.
Restored the standard upgrade dialog for blocked video lectures, podcasts, and slides.
Upgrade prompts now use bright gold in light and dark mode.
Repeated failed sign-ins now pause briefly, and the login page offers an emailed sign-in link so you can get straight back in.
Project invitees now see a project's name, description, and instructions as read-only, and shared links' passwords, expiry, and download settings now apply to exports too.
Asking chat to remake a video lecture with a fix, such as a corrected pronunciation, now makes the new version instead of asking for the original video file.
Study guides and worksheets in Traditional Chinese now print every character.
A worksheet or study guide whose build failed now gets rebuilt, instead of retrying the same check for most of an hour.
Chat can rename an uploaded PDF, and a rename or move it cannot do no longer stops the reply partway.
Creating something from a PDF that is still being read after upload now waits for it, instead of saying the PDF has no text.
Deleting a meeting recording right after an app update no longer fails with "Failed to delete this recording".
Scholarly works again in older browsers that were missing two features chat and uploads rely on.
Devices that go unused for six months are now signed out automatically.
Japanese and Chinese narration now reads technical terms, kanji, and math more accurately and sounds more natural.
Narration in Belarusian, Catalan, Filipino, Greek, Hebrew, Malay, Persian, Polish, Serbian, and Slovenian now uses voices that pronounce words and numbers more reliably.
Read aloud and chat voice notes now use the narration voice that reads each language best.
Video styles are listed by name and description again when you create a video lecture, on desktop and phones.
Checkout is simpler: it asks only for your payment details, with no business name or tax ID fields.

November 4, 2026

Create and work by voice

Ask voice to create study materials, use your connected apps, or work with your files in the same conversation.
Voice can discover tools and use your saved skills, preferences, and memories.
Finished tool results stay in chat when you interrupt or reconnect, and queued actions stop cleanly.

A visual video style picker

Video styles now appear as large example tiles on desktop, so you can choose by look.
On desktop, press play on any style to watch a short example, then pick it from the preview.
Video creation on phones opens with a compact style selector, keeping settings and Create within easy reach.

Smaller changes

Some phone accounts can email themselves a sign-in link to continue on a computer, or keep using their phone.
Some phone accounts see a simpler Design dropdown when choosing a video style.
Some accounts see a compact video style list while we compare it with example previews.
Lecture-ready email links can reopen their upgrade offer, and waiting lectures stay visible when you switch accounts.
Some Free accounts see brighter gold upgrade prompts while we compare the two existing color styles.
Restored the standard locked video-length selector for Free accounts.
Restored the standard file-size, export and recording-limit upgrade messages for Free accounts.
The "your lecture is ready" email now lists what the full-length version adds, with a link to make it longer.
Long video lectures no longer fail while being saved at the end.
Edits now save, and recording now starts, in browsers that block site data.
Study guides and worksheets keep their equations formatted when one contains an accented letter.
Documents that cannot be finished now stop in minutes with an error instead of running for an hour.
Study guides and worksheets from Sinhala, Tamil, and other sources we can't typeset yet are written in English instead of failing.
A chat task that runs out of time now says so instead of reporting an error.
Saved preferences now show in Custom instructions as chips you can remove with one tap.
Ultimate shows its credit cost right in the style picker, before you choose it.
Settings splits AI & Chat into Personalization, Memory, AI Creations, and Skills, so each is easier to find.
Narrate as written no longer skips parts of a script that is too long for the video. It stops before starting and says how to split it or pick a longer length.
Freestyle and Illustrated videos reveal each highlighted item when the narration reaches it, instead of ahead of the voice.

November 3, 2026

Six video styles, with Shorts for every one

Choose Standard, Standard Fast, Illustrated, Math Lecture, Doodles, or Freestyle.
Turn on Short video for a focused portrait version in any style.
Compare example thumbnails, then click or tap one for a closer look.
The new Illustrated style pairs rich artwork with lively narration and paper-strip captions.

Smaller changes

Video lectures better preserve supplied narration scripts, documentary-style visual requests, and requested math formulas and labels.
Improved documentary visuals in long video lectures built from text sources.
Ultimate remains available to Laureate members in the new video picker, including when reopening saved settings.
See what an expanded slide deck, more flashcards, or continued quiz practice adds before upgrading, then pick up where you left off.
New chats open directly to the composer, without generic starter questions or an upload prompt.
Short podcasts and video lectures now preview the source topics a longer version would add before you upgrade.
Office uploads can finish larger documents without short rendering cutoffs and clean up interrupted conversions promptly.
Restored standard Free upload, AI creation, and recording limit messages.
Free video lectures use standard processing again to reduce generation waits.
Large PDFs extract text faster, use less preview memory, and recover more reliably from interrupted downloads and rendering failures.
Restored standard processing for summaries, flashcards, and slide preparation to reduce generation waits.
Any File now delivers requested editable PowerPoint decks without getting stuck retrying or requiring a switch to Word.
Word files containing only images or equations now finish without being mistaken for empty documents.
Any File can now create native .doc, .ppt, and .xls files for older versions of Microsoft Office.

November 2, 2026

24 more languages across Scholarly

Use Scholarly in 24 more languages, including European Portuguese, Greek, Hebrew, the Nordic and Baltic languages, Polish, Czech, Hungarian, Romanian, Slovak, and Traditional Chinese.
Hebrew uses right-to-left navigation throughout the interface.
Read the website and help center in your language, and choose your interface language in Settings → Appearance.
Localized website and help pages keep their light appearance while loading.

Smaller changes

Checkout no longer gets stuck if your browser cancels saving an upload selection.
Some Free users see clearer upgrade copy for their next upload, longer video lectures, and more AI creations.
Some Free users now see clearer recording upgrade copy explaining the longer limit for new lectures.
The lecture recorder page is testing a clearer invitation to turn a recording into notes.
Some Free users see clearer Premium offers for their selected export, header upgrade button, and future recordings after a duration limit.
Failed-payment emails and billing notices now explain how to try another card or payment method, including after a Link payment fails.
Ultimate video mode for Laureate now costs 10 AI Creation Credits, down from 15.
Free video Auto no longer selects a model that caused unusually long generation waits.
Reading a long saved agent note keeps emoji intact so it cannot interrupt generation.
Subscription reactivation and yearly billing now let you complete bank authentication without losing your purchase or changing your plan before payment.
Finished recordings no longer show a separate recap offer above their creation actions.
Unfinished yearly upgrades now offer a monthly option by email, with help trying another payment method.
Some Free accounts can try a larger lifetime AI Chat allowance, with their assigned limit shown in chat and Billing.

November 1, 2026

Ultimate video mode

Ultimate mode is in beta, appears only for Laureate members, and uses more AI compute for a better video result.
Each video costs 15 AI Creation Credits, shown clearly before you create it.

Smaller changes

Reactivating a subscription no longer risks a duplicate bill, and failed payments keep your existing plan unchanged.
Your personal downgrade and Undo controls stay available when you join a team, and Enterprise accounts only show subscription controls when applicable.
Tell any creator what you want from now on, like "always keep sentences short", and it applies to your later creations. You can see and remove these in the create step and in Settings.
Video, document, and research agents can now keep helpers and notes for your later runs. See or delete them in Settings under Saved by agents.
Laureate members can choose Downgrade to Premium from the plan picker and review the change in Billing.
Free video lectures now honor a model you pick explicitly; Auto lets Scholarly choose the model.
The sidebar is tidier: Folders appears once a folder has something in it, and new folders start in Library.
Sidebar folders show their real item count and load older items when opened.
Long titles no longer squeeze sidebar folder icons.
Voice mode no longer ends with an error when you talk over a reply just as it finishes.
Figures taken from a PDF in video lectures and slides now come from the right page.
A file we truly can't read now stops with a clear message before any credit is used, on every creator, naming the file. With a prompt, we make it from your prompt and tell you which file was left out. Short or scanned sources keep working.
Gemini 3.8 Flash is no longer offered for video lectures; a saved choice now uses Auto.
Free video lectures now use a lower-cost processing tier and may take longer to finish.
Voice mode now ends itself after a quiet minute, and the Free plan includes 5 lifetime voice minutes.
Standard voice previews now play in your video or podcast's language when you pick Spanish, Portuguese, Japanese, and 25 others.
Videos and podcasts in regional languages like Spanish (Spain), Brazilian Portuguese, and Canadian French get new default voices. A Google voice still speaks the variant by name, for one extra credit you see before you pick it.
A podcast that comes in shorter than its length tier now returns the extra length credits, the same as a video.
Close the tab on a recording and it still finishes: a day after the last audio, a recording of a minute or more gets its transcript and notes automatically.
A chat turn no longer fails when a source contains an emoji or other special character that landed on the edge of a trimmed excerpt.
On the free tool pages, closing the file picker without choosing a file no longer leaves the upload button doing nothing.
Opening an email sign-in link a second time now takes you to your content instead of back to the login page.

October 31, 2026

Word, PowerPoint, and cloud files work like PDFs

Word, PowerPoint, OpenDocument, and RTF files keep their original file and open like a PDF, with pages, a summary, and page citations.
Pick pages or chapters from these files in every creator, and chat reads and cites them by page.
Add files from Google Drive, Dropbox, and Notion in every creator, in chat, and in Deep Research.
Notes, tables, and pictures now open in the same page layout.

Build, explore, and share 3D models

Attach a PDF in chat and ask for a 3D model, or accept the 3D offer, to build from that document and explore the result in chat.
A first build appears as soon as it finishes, and a new model from a PDF suggests what to build.
Move live sliders and switches while a model runs, then reset them to try again.
Saved pictures show your models in the library, in chat on phones, and while a model loads.
Public links open without an account, or you can restrict access to invited people or your team. The source document keeps its own access settings.

Every creator takes every source

3D models, slides, documents, study guides, mind maps, infographics, essays, read-aloud, flashcards, quizzes, and Image Studio now take the same files as video lectures and podcasts.
Build from Word and PowerPoint files, videos, audio, recordings, podcasts, research reports, images, links, Google Drive, Dropbox, and Notion.
3D models can use several sources at once, pick pages, or start from a typed description.

Smaller changes

The page you see when a shared item is private is cleaner: what it is and who shared it sit in one box, with the request and sign-in options right under it.
Some signed-out visitors to shared content can create an account directly over the item, resume after reloading, and return to it after signing up without losing a selected plan.
Flashcards and quizzes made from selected chapters now stay inside those chapters, instead of adding cards from the chapters between them.
Changing a queued creation's time now moves later creations with it even when an item in the middle was just canceled.
For now, the chat beside a 3D model runs on GPT 6 Luna or Claude Haiku, so its model menu lists just those two.
Clearing a 3D model also removes its saved preview picture.
Exported 3D model images and videos now match what you see on screen, instead of showing parts lit only by the default light as black.
A 3D turntable video now runs its full 8 seconds and starts from the first angle, instead of coming out short when the model took a moment to load.
Generated 3D models are checked for labels that cover each other before they reach you.
A picture you add to a study guide now appears in it.
Slide text that is too long for its space now ends on a finished sentence where it can, instead of trailing off.
Audio made from a Word file is now labeled as audio in your Library, instead of keeping the .docx name.
A finished research report can go straight into any creator with Make from this.
Connecting Google Drive, Dropbox, or Notion now brings you back with everything intact: sources and page choices while creating something, your upload selection, your research question, and your unsent chat message and attachments. The same goes if you cancel or the connection fails.
On a phone, the large Play button on a video now plays it wherever you tap it.
The page-range slider now works from the keyboard.
A damaged Word or PowerPoint file now gets a clear message and keeps your upload, instead of opening as pages of unreadable text.
Word and PowerPoint pages no longer call the document a PDF in chat suggestions, and a 3D model's name stays readable on phones.
Read aloud works again. The voice service changed the shape of its audio and every request was rejected; read aloud now accepts it, and falls back to generating the whole clip if a live stream fails before any sound arrives.
Starting a meeting recording no longer gets stuck on "Another AI Meeting Notes capture is already in progress" when no other tab is actually recording.

October 30, 2026

A Tasks page for everything in progress

Tasks, in the sidebar next to Library, lists everything running, scheduled, repeating, or needing a look, plus what finished in the last week.
Artifact Tasks like video lectures, flashcards, and slides sit beside Chat Tasks, with a search and a filter for each.
Start, move, or cancel anything waiting right from its row, and retry or dismiss what failed. A count on the sidebar shows what is running or failed.
On paid plans, ask chat to repeat a task every day, week, or month for up to 12 runs, with results posted in the same chat.
Repeats run at the same local time across daylight saving changes. Home shows each task once, with its schedule and remaining runs.

Make more from your recordings

Video lectures can use a recording's full transcript, notes, and topic timestamps.
Follow-up actions adapt to lectures, meetings, and training, with a choice to change the purpose.
Some recordings now offer a quick review with saved answers and replay links to the supporting moments, without using an AI Creation Credit.

A cleaner, easier-to-read Scholarly

A new interface typeface with real medium weights, softer consistent corners, and lighter icons.
A clearer dark mode, with brighter text and surfaces that separate by shade instead of outlines.
Tidier lists, chat cards, study pages, Settings, and Team pages, with secondary buttons that stay visible next to the main one.
Image Studio now opens on a prompt, with example images to start from and your recent documents one click away from a set of diagrams and illustrations.

Smaller changes

Stop every remaining repeat from its chat card. Repeats also stop on their own after three results in a row go unopened.
The AI Chat model menu is about half as tall: search stays up front, while Unlimited, All models, and Thinking each open beside it.
Claude Haiku 5.5 in AI Chat is now unlimited on paid plans and no longer uses premium usage.
Your premium usage now lives in Settings under Usage, one click from the AI Chat model menu.
Weekly premium chat allowances are 25% lower on Premium Lite and one-third lower on Premium and Laureate. Chat continues on Auto after the allowance runs out.
Removed retired community links from the app, website, and emails. Help and email support remain available.
Ask for a 3D model in any chat: it builds while you keep working, then chat replies with a model you can spin, zoom into, or open full page.
Video lectures in Afrikaans and other languages our standard voices can't speak now get full narration instead of ending early.
A very long chat thread opens and keeps working again. Past a certain size the thread's own saved copy could no longer be stored, and anything that read the whole thread failed instead of just reading it the slower way.
A creation no longer fails because its progress message thread became unavailable. The creation keeps going without the thread instead of ending partway through and spending your credit.
Finishing a long recording no longer fails over a single storage hiccup: each piece of audio is now fetched on its own and retried, instead of one slow read discarding the whole recording.
Audio we cannot transcribe now says so right away, and says why, instead of retrying for several minutes and then asking you to try again. A creation that uses that audio stops instead of making three more attempts that cannot succeed.
The free creation offered while canceling is back for everyone with a source to build from. Since October 20 it was hidden from anyone who still had most of the week's creations left, which is almost everyone.
3D models build again. Since October 28 a build could run until its time limit and then fail, returning your credit.
A long-running command in AI Chat now stops at the time limit with its output and a note, and the reply carries on, instead of ending the reply with a warning that the command may have completed.
A recording whose audio timing jumps partway through is now transcribed in full, instead of being cut short and reported as having no speech.
Safari now keeps your selected file when you upgrade from the upload window, instead of asking you to choose it again.
Flashcards from a YouTube video with no speech now say so, instead of asking you to wait for a transcript that never arrives.
The citation generator now has a Cite another source button, so you can keep adding to your bibliography without reloading.
The math solver now shows every step by default.
Opening a page, deck, PDF, or file a repeating chat task saved now counts as reading its result, so the task keeps repeating.
Asking chat to repeat a task on Free now shows an upgrade card in the chat, and chat can still schedule it to run once.
A repeating quiz, flashcard set, or notes summary from chat now saves to your library on every run, so each one is there to study later.
Scheduling a creation now starts with nothing picked: choose This evening, Tomorrow morning, or your own date and time. A start less than 15 minutes away offers Create now instead.
Queue a creation to start when your current one finishes, with After my current creation in the schedule options. Home shows what it is waiting for, and it starts anyway if that creation runs much longer than usual.
Scheduled creations on Home now have Change time and Cancel next to Run now, and a repeating chat task can stop all its remaining runs from Home. Both ask before cancelling.
On a phone, tap a scheduled creation or chat task on Home to see when it runs and how many runs are left, start it now, change its time, or cancel it.
Scheduled flashcards, quizzes, and research can now be cancelled from Home before they start, and the credit comes back.
Cancelling a scheduled document, slide deck, infographic, or mind map before it starts now gives its credit back, like every other creation.
A scheduled creation's chat card, task page, research page, and project row now show its current time and let you run it now, change its time, or cancel it. A card for a creation you cancelled elsewhere now says so.
Moving a scheduled creation now moves anything queued behind it, and cancelling one in the middle of a queue no longer starts the next one early.
A repeating chat task no longer stops itself right after you change its time or run it early. If your plan ended the repeats, its card now offers the way back.
You can now ask chat to move a scheduled creation or chat task, from any chat.
Chat tasks now have Change time, on Home and on their card in the chat. Moving a repeating task's next run moves the later runs with it, keeping the same spacing, and you can also ask chat to move it.

October 29, 2026

Cast your podcast, and pick from 66 voices

A new Cast section names your host, guest, and producer, gives each one a voice, and remembers the cast for your next episode.
Tell us who the guest is, like a cardiologist, and the host welcomes them by name on air. Turn the producer off for a two-person show.
Google and Grok voices join the standard voices for podcasts, video lectures, and read-as-MP3 audio, for one extra credit.
Give any finished podcast, video lecture, or audio file new voices without rewriting it. Chat can do it for a podcast or video lecture, and the progress page shows only the recording steps.
Every voice has its own character, and you can favorite the ones you like or shuffle a fresh cast.

Ask for changes from the start

Send requests while podcasts, videos, slides, and visual documents are queued or scheduled. Scholarly replies when the worker is ready.
Redirect a podcast's script before narration begins, and guide video and slide planning from the same thread.
See whether your request was read, applied, answered, or could not be made.

Real pictures in chat, in a swipeable row

Chat shows real photos, labeled diagrams, maps, and artworks on its own whenever a picture helps the answer, without you asking.
Several pictures sit in a row across the chat, each at its own shape: swipe it on your phone or use the arrows on desktop.
Tap a picture to open the set full screen, then swipe or use the arrow keys to move through it, with who made each picture and where it comes from.
Charts and AI images chat makes sit in one row under its answer, with a +N tile for the rest; chat can point to one mid-sentence with a small chip that opens it.

Smaller changes

Free includes one lifetime recording of up to 60 minutes, with transcript and notes included. Saved recordings remain available.
Oversized uploads now consistently show the file-size limit and upgrade options, without experimental shrink or first-pages offers.
More languages now offer the standard narration voice library; regional language variants keep their dedicated voices.
Video lectures with still images now keep the picture on screen for the full narration instead of failing near the end of generation.
Recordings now get a full transcription pass after capture for a more accurate final transcript, while live previews remain available during recording.
Meetings show accurate transcription progress, accessible playback and menu controls, and the meeting allowance in upgrade prompts.
Signups bringing a file for a video lecture now consistently use the standard welcome instead of an experimental plan-choice screen.
New free accounts are asked what their material is for before their first creation, and the answer suggests a format to start with.
New desktop accounts start on one welcome screen: drop a file to make a video lecture, or start from a topic, link, recording, or chat.
Shortened free video lectures no longer list the sections they left out; the option to make the full lecture is unchanged.
Every creator checks your sources the same way before it starts: a blank PDF or a saved viewer page is refused on the create screen with the reason, instead of failing after you wait.
A creation that hits a temporary problem now says it is retrying automatically, with the attempt number, on its progress page, Home, and in chat, instead of looking unchanged until it finishes.

October 28, 2026

Pin messages in chat

Pin any answer or your own message from its hover actions, the right-click menu, or a long press on your phone.
Pinned messages sit in a bar at the top of the chat: tap it to jump to each pin, or open the list to see them all.
Ask chat to pin something, or tap Pin when it offers to keep a deadline, plan, or key fact for later.

Chat tasks file into your project or folder

A task you hand to chat from a project or folder, now or scheduled for later, saves its result into that project or folder.
Name a different project or folder in the request and chat files the result there instead.
The task card shows where the result will land, and chat confirms it when the task is set up.

Ask for a change while it's being made

Agent-built video lectures, slides, documents, and research have a thread with their agent on the progress page; podcast threads cover extras after the episode script is written.
The agent posts its plan once it has one and a line when each phase starts, so the page is never blank while you wait.
Type a change or a question into Ask for a change; the agent reads it between steps, applies it when it can, and replies there, with Sent turning to Seen.
Chat can relay a request into a running creation for you, shown in the thread as From chat.
A running creation's row on Home shows the agent's latest note. Each creation takes up to 15 requests.

Smaller changes

Page PDF exports now typeset inline equations, standalone equations, and math in tables; Summary PDFs keep headings and checklist items with their content.
Slide practice includes visible questions, while flashcard and lecture reviews check answer clues and signed arithmetic more explicitly.
Exam feedback removes stray control escapes, autocomplete stays grounded in your notes, and podcast summaries focus on the topic instead of presenter names.
Resuming an annual upgrade keeps annual billing selected, and longer-version offers no longer describe the plan limit as your recording's exact length.
Recording the same meeting in two browser tabs no longer starts competing captures, and opening a new chat waits for it to be ready.
The lab report writer checks for blank required fields before sending your request.
Free video lectures spend their two minutes finishing one example, story or explanation in the style you asked for, and the Upgrade button under the video is now blue.
Credit packs now come in four sizes: Small (10 for $10), Medium (35 for $30), Large (110 for $90), and Heavy (250 for $200), with bonus credits shown on the larger packs.
Fixed file and recording uploads failing after the API startup update.
A question sent from Home is no longer lost when the app reloads for an update on its way to chat.
If our PDF reader has a problem, an upload now asks you to try again in a moment instead of calling your file corrupted.
The progress page shows a stage bar with the stage names, the current step and its usual time, and Cancel beside the title.
The Design step when creating a video, infographic, mind map, or story book is now a list that shows every design with its description, instead of a row of uneven tiles.
Fixed study guides, worksheets, and cheat sheets failing to generate when a heading, table cell, or defined term ran onto a second line.
Chat no longer stops mid-answer when one of its commands runs for more than about three minutes, or when a file it makes is over your plan's size limit; it carries on and works around it instead.
A chat message the server refused because its id was already taken can now be sent with Retry instead of failing every time.
A new flashcard deck now opens with its quiz already written for every card, so Quiz starts instantly. The progress page shows a Quiz step while the questions are written.
A chat answer whose connection dropped mid-reply now keeps streaming in place, and Refresh picks it back up, instead of saying it is still finishing until the reply ends.
A chat message sent while the server is briefly busy is retried for you instead of failing.
A page keeps its own language when the first download of its translations is interrupted, instead of falling back to English.
Fixed video lectures failing when a reused figure or picture was saved to storage twice in the same second.
Opening a question during a podcast now pauses the episode until you are done with it, the way it already works for video lectures.
The link button in the notes editor now announces itself to screen readers.
Some new sign-ups now see the video or podcast they came to make on the sign-up page, then start on one screen for it: add the file and it starts right away, with your plan's limits shown up front.
On paid plans, every named model in AI Chat now uses your weekly premium usage, with lighter models such as Claude Haiku 5.5 or DeepSeek V4.1 Flash using it at the slowest rate. Auto has a separate weekly fair-use limit.
The AI Chat model menu has a Thinking slider (Fast, Balanced or Deep) and lists every model in one searchable list grouped by lab.
Paid members with an older saved chat model start on Best once; a new model choice stays saved.
Free plans chat on Auto. The model menu still lists every model, and choosing one shows the plans that include it.
AI Chat adds Gemini 4 Argon, Muse Wonder, and Grok 4.7. Mistral Small 4 is retired, and Gemini 3.8 Flash, Gemini 3 Flash, Gemini 3.5 Flash Lite, and GLM 5.3 Flash leave chat but stay available when creating.
Compare Scholarly with Grok, Claude, and Microsoft Copilot on three new comparison pages.
Logging in with an email that has no account offers Create account right under the field and catches typos like gmial.com; a Google account shows Continue with Google instead of a password field, and a wrong password shows Reset password beside it.
When the microphone is blocked, recording, dictation, and Talk show the steps to allow it in your browser and try again on their own once you do.
Finishing a recording retries on its own through brief connection drops, and processing a deleted meeting restores it first.
Video lectures in progress have Edit & restart, which cancels the run and reopens it with the same sources and settings.
Opening someone else's progress link now says it belongs to someone else instead of showing an error.
Files that aren't downloaded to your device yet (iCloud, OneDrive, or Google Drive) are flagged as soon as you add them, and uploads retry on their own through brief connection drops.
Connecting Google Drive shows which box to tick before Google's permission screen, and a cancelled connection offers Try again.
Videos and podcasts that fail to load try a fresh copy before showing an error, and flashcard Auto Play reads a card with your browser's voice instead of stopping when ours can't.
Emailing yourself a link to a page now goes through when you try again, instead of failing for the rest of the day.

October 27, 2026

Choose pages or chapters

Pick the source content for videos, flashcards, quizzes, podcasts, slides, and other creations.
Switch between pages and chapters without losing your selection.
Browse all chapters for free, with chapter selection available on Premium.

A cleaner, more complete help center

New guides for listening to a lecture, podcast feeds and embeds, making a lecture in another language, and seeing who on your team finished a lecture.
A new guide explains how Scholarly reviews what it makes against your sources before you see it.
Study Guides are now called Summaries throughout the help center, matching the app.
Seven guides that repeated another guide under a different title were merged into the one that owns the topic. Old links still work.
The Notion and Dropbox connections and the redesigned progress page are now documented, and the free quiz pool is described correctly.

Smaller changes

While a creation renders on the Free plan, the task page shows your progress first with one short upgrade line below it.
Signing up from a tool page takes you straight to that creator with your file attached and ready to create.
Menus open reliably when their buttons finish loading after the surrounding page.
The Getting Started guide is shorter and easier to follow, and answers what a creation costs, why one takes a few minutes, and how to get it in another language.
Video lecture and podcast pages now describe listening instead of watching, re-narration in another language, private podcast feeds, and embedding.
The Help Center assistant matches in-app chat, and Help and marketing pages stay in light mode.
Chat now answers more warmly and does more of the work in the same reply: plans, schedules, and explanations arrive right away, and it fixes its own mistakes instead of describing them.
Chat's Best mode now uses GPT 6.1 Sol for stronger answers. GPT 6.1 Terra remains available in the model picker.
Planning with chat uploads Word documents and other selected files before starting the conversation, keeping your request and selected sources intact when an upload needs a retry.
Retrying a stopped chat follow-up preserves the interrupted message reference.
Recordings can finish after a device sleeps overnight by transcribing the saved audio when live preview timing is unusable.
Emailing a saved link works after its title or your interface language changes.
Choose PDF pages for every new video lecture, with the upgrade option visible when you need more pages.
The homepage opens with the interactive composer so you can explore different ways to use your material.
Task updates stop cleanly when you leave or switch accounts, and Explore refreshes recover more reliably after visibility changes or slow queue setup.
Shared pages no longer flash "Reconnecting" or drop collaborators when you switch back to the tab.
When you ask chat about your premium usage, it now lists GPT 6.1 Terra among the models that use it.
Sending a chat message again after it fails now works even if you pick another model or change a setting first.
Pressing "Email it to me" twice no longer reports an error for an email that was already sent.
Video lectures that reuse the same figure or picture across scenes no longer fail while saving their progress.
The header Upgrade button shows how many free AI creations you have left for everyone on the Free plan.
Upgrade offers use a softer gold throughout the app.
Limit walls return to a single See plans action; the plan comparison opens before any payment step.
The file-too-large notice is back to its earlier red alert with a one-tap Upgrade.
Plan limits read faster: each one names what stopped, what your plan allows and what the next plan adds, with a single Upgrade button.

October 26, 2026

Italian and Turkish

Use Italian or Turkish throughout the dashboard, marketing pages, and Help Center.
Choose Italiano or Türkçe in Appearance settings or the public language menu. New accounts can follow their browser language.

Scholarly in Arabic

Use Arabic throughout the dashboard, marketing pages, and Help Center.
Navigation, forms, and help articles follow a right-to-left layout, with code keeping its own reading direction.
Choose Arabic in Appearance settings or from the public language menu. New accounts can follow Arabic browser preferences.

Standard Fast video lectures

New videos start on Standard Fast: designed slides with fewer effects and fewer authoring passes. Standard remains available.
Long lectures check slide corrections before encoding and reuse unchanged parts during repairs.
Images and narration can run at the same time, and video startup reuses rendering assets already installed.

Smaller changes

Video wait estimates use recent completion times for each style and length, including Standard Fast, Illustrated, Freestyle, and Math.
Video slide decks export reliably when the narration ends just after the final video frame.
Switching a lecture to audio preserves mute and volume, with an unmute control in the audio player.
Slide deck corrections keep the right image on each repaired page while reusing unchanged slides.
Signup keeps the language of the public page you came from, even before the page finishes loading.
Best creations use more efficient model handoffs for video slides, summaries, essays, podcast extras, visual checks, and research workers.
Video and slide comparisons keep their explanations when three or more concepts are shown as cards.
Adjusted weekly premium chat allowances: Premium provides 3× Lite's allowance and Laureate provides 4× Premium's. Auto chat remains unlimited on paid plans.
Compare Premium Lite and Premium side by side in eligible countries, with Laureate below and simpler plan descriptions.
Practice questions preserve the setup needed to answer them, and study guides and worksheets review missing answers and mismatched instructions.
Chat's Best mode now uses GPT 6.1 Terra to stretch your weekly premium usage further. GPT 6.1 Sol remains available in the model picker.
Podcasts check continuity across the episode, flashcards review missing context, and slides check examples and complete answers against their speaker notes.
PDF study guides check that review topics have been explained before asking you to study them.
Generation status updates arrive sooner, and repeated source and image searches share their work while Explore refreshes in the background.
Video generation timing now measures a retry from when it was restarted.
Localized navigation, pricing controls, help links, and creator options fit more reliably on phones and in right-to-left layouts.
Standard video lectures now edit narration with the whole lesson in view, making topic changes, examples, and returns to earlier ideas easier to follow.
Invite editors to shared pages and flashcard decks, receive live updates, and recover conflicting drafts during the collaboration rollout.
Collaborator presence stays consistent across realtime servers and clears automatically after a disconnect.

October 25, 2026

Chat answers with the best model

Premium and Laureate chats now start on Best, our strongest model. Auto is one tap away for fast, unlimited answers.
Claude Opus, Claude Sonnet and GPT Sol are now available in chat on Premium and Laureate.
A weekly premium usage meter in the model menu shows how much is left and when it resets. Laureate gets five times more.
If you use it all, chat moves to Auto on its own and returns to Best after the reset.
Background tasks that chat starts, like building a file or doing research, run on the same model as your conversation.

Marketing and help in your language

Browse marketing and help pages in the same 11 interface languages as the dashboard.
Find updated help for current features, creation flows, plans, and account settings.

Smaller changes

Video subtitles now use the narration language to help distinguish Hindi from Urdu.
Long video lectures whose render starts late in the run now finish on the first try, instead of stalling for up to two hours before a retry.
Video generation starts faster by reusing installed rendering tools.
Video and slide previews reuse unchanged slides, and video repairs appear before encoding.
Video lectures render faster, and fixing a scene after review no longer re-encodes the whole video.
Chat background tasks start faster and load files and images more efficiently.
Image-heavy chats respond faster by reusing uploaded and prepared images.
Chat starts working with downloaded source files sooner.
Blog posts, help articles, and public pricing facts now quote Premium at $180 a year, which is $15 a month and 50% less than paying monthly.
Small generated files and previews upload faster.

October 24, 2026

Chat reads your whole source and shows where it found the answer

Scanned PDFs, image-only pages and tables are read from the page itself, so answers no longer describe a watermark or miss a rubric.
When a long video, recording or transcript is only partly read, chat says so, and it reads the whole thing when you ask for full notes.
Hover a citation, or tap it on a phone, to see the exact passage and page before you open it. Open page or Play from jumps straight there.

Creating from chat just works

On Free, a video or podcast up to 2 minutes starts straight away; anything longer shows its upgrade option once instead of asking again and again.
Asking again while something is already being made points you to it, with no second credit, and chat can tell you how long it usually takes.
Files you drop into chat and video transcripts can be used to make study guides, documents and visuals.
When you say you are saving credits, chat tells you the cost and waits for your yes.

Change what Scholarly made, right in chat

Saving an answer as a page no longer uses an AI Creation Credit.
Reply "yes" or "accept" to apply a suggested edit, and undo an applied edit for 30 days. Tables can be edited too.
Ask for a new version of a video lecture in another language, length or style, and it is made as a new video from the same material.
Length, logo and language you type in chat now win over the create window's defaults.

Smaller changes

Finished essays remain available if their final editing pass times out.
Video generation can inspect newly rendered contact sheets immediately, reducing extra review steps.
The upgrade window has a larger, square close button and a simpler Premium credit row.
Laureate stays visible alongside the other plans on pricing pages and in the upgrade window.
Video lecture and podcast scripts get an editing pass before they are voiced, so narration sounds more natural: clearer sentences, smoother transitions and less filler, with the same facts and length.
Claude Opus 5.5 now costs 2 AI Creation Credits instead of 3, and Use best model now picks Claude Opus 5.5.
Every flashcard deck now gets a final edit that checks each card against your source and makes questions clearer, answers tighter and hints more useful. With Claude Opus 5.5 or Use best model, GPT 6 Luna writes the deck and Opus does the edit.
After you upgrade from a page, file-size, or upload limit, Scholarly reopens what you were making, whether a video lecture, podcast, slides, flashcards, quiz, summary, essay or another creation, with your file and page range (and your settings for videos, podcasts and slides), one tap from Create.
Standard video lectures finish visual corrections faster.
Pages keep their translated text when an update rolls out, without downloading the same translations again while opening.
Video lectures finish audio processing sooner while keeping speech-aligned captions.
Video lectures review generated artwork as soon as it arrives, avoiding an extra inspection step.
Essays, study guides, story books and infographics get a final edit that checks them against your sources and tightens the writing before you see them.
Standard video lectures preview their slides before encoding, making visual corrections faster.
Standard video lectures reuse clear prepared narration and adjust pacing before recording, reducing unnecessary rewrites.
Videos assembled from still images render faster, especially longer Illustrated and Freestyle lectures.
Some visitors to the Notes to Video and PDF to Video pages now see a sample video, what happens after signup, and how credits and plan access work, and sign-up buttons on both pages go straight to registration.
Expert styles are back as cards with their descriptions in the video, podcast, slides, study guide, flashcard, quiz and Story Book create windows, above Settings, instead of a menu under Advanced options.
Premium Lite is available in selected regions at $12 a month or $72 a year, with 5 AI Creation Credits a week and 50% savings on yearly billing.
New Premium subscriptions cost $30 monthly or $180 yearly, with rounded local prices; Laureate local prices now use rounded $99 equivalents. Existing subscriptions keep their current price.
The Share window on a lecture or podcast has two new tabs: Podcast feed puts all your podcasts and lecture audio in Apple Podcasts or another podcast app through a private link, and Embed gives you code to play a shared lecture in Moodle, Canvas, Notion, Confluence or SharePoint.
Private podcast links clear when you switch accounts, and welcome actions wait for your file or plan handoff to finish.
Lecture progress shows translated creation suggestions and waits for its upgrade layout before displaying it.
Real image searches keep distinct photos with the same title and return the requested number when more images are available.
Chat keeps answers visible across repeated tool lookups, passes every linked source to helpers, and preserves citation examples inside code.
A free lecture's section list follows its scene sources, and longer-version offers stay within the selected PDF pages.
Chat now sees which flashcards you get wrong, so "which cards do I keep missing?" gets a real answer, and the empty chat page offers to pick up your most recent conversation.
Chat replies in the language you write in, even when your app language is different.
Chat no longer narrates its own steps or tool problems, and it asks at most one question before getting started.
Chat only says something is saved, updated or ready when it actually is, and quotes and page numbers come from what it read.
Long chat turns always end with an answer, and repeated identical searches no longer slow a reply down.
Questions about your plan, credits and limits are answered from your account.
Math video lectures come much closer to the length you pick, and right-to-left lectures turn pages in the right direction.
Podcasts made with a premium model are now planned by that model and written to its plan, so they explain with clearer examples and stay more accurate.
Video lectures and slide decks made with a premium model on Auto language now stay in your source's language instead of sometimes switching to another one.
While a video lecture renders, its progress page says how many sections it plans and how long it runs, and a Free lecture shows which sections of your PDF it covers and which the full lecture would add.
In voice mode, a lecture you are recording is kept as one passage and you are offered Record; asking voice to make something leaves it in the chat box for you to send.
The "Send this lecture to your team" row is gone from the lecture page, so the area under the player is simpler. Links you already sent keep working.
Slide decks get a final edit of every headline, label and table on screen, so each slide states its point clearly; with a premium model, that model does the edit.

October 23, 2026

Real images in chat

Ask to see something real, like a nephron, a painting, or the Berlin Wall in 1989, and chat shows freely licensed photos and diagrams from Wikipedia and Wikimedia Commons instead of an AI illustration.
Several images sit together as one strip; open any of them full size to see who made it, its license, and a link to the source.
Diagrams stay readable in dark mode, and answers in other languages prefer images labeled in that language.

Listen to a lecture instead, or send it to your phone

Every video lecture and uploaded video has Listen instead under the player: the same narration as audio, with chapters, your speed and your place, and it keeps playing with the screen locked.
On desktop, Send to my phone next to Share shows a QR code, emails the link to your own address, or copies it. The link opens the lecture ready to listen.
The lecture-ready email and the finished lecture on Home have a Listen link too.

See who finished a lecture shared with your team

On a paid Scholarly Teams plan, a video lecture or podcast shared with your team shows each teammate who opened it, how far they watched, whether they finished, and how they did on its questions.
Open it from See who finished in the Share window, or from the new Shared lectures section in Team activity. Owners and Admins can download the table as CSV.
Send to your team: email a lecture you made from a work document to up to 10 colleagues, each with their own link and no account needed, and see how many opened and finished it.

Smaller changes

Someone watching a lecture you shared can drop their own PDF, slides or notes above the player to make one like it, and nothing pops up over the lecture while it plays.
The Made with Scholarly mark on a shared player opens in a new tab without pausing the lecture.
Signed-out viewers can answer every question in a shared lecture; their answers come with them when they sign up, and they see their score.
Opening a private item now shows a preview, and asking for access takes only an email. The owner can let them view with one tap from the request email.
Questions and the end card now appear over a fullscreen lecture.
When a creation finishes while its tab is in the background, the tab title shows ✓ Ready and the icon turns into a green check until you come back.
Creation-ready emails open your creation already signed in on the device you read them on. Each link works once, for 24 hours.
New video lectures show which page of your PDF the scene on screen came from, under the player and on each transcript line.
Click a page to open your PDF at that page over the lecture, or open the full document in a new tab.
If a lecture comes out narrated in a language you don't read, its page offers one free remake in your language, with the same source, length and style.
Make a finished lecture in another language for one AI Creation Credit: the same scenes and figures, narrated and labeled natively, with every language version linked under Also in.
Text, Markdown, CSV and other document files are no longer capped at 10 MB when you make a video, podcast, slides, study guide, flashcards or other study material, or ask about them in chat. Any file your plan lets you upload can be used.
Very long files are read the way long PDFs are: generation uses up to about 1,000 pages' worth of text, and video lectures, documents, slides and chat can open the complete file for anything further in.
Audio files in your library, such as narrations and recorded lectures, can now be a source for any creation or chat. Scholarly uses their transcript and starts one if needed.
A PDF saved under another name, such as a download Chrome never finished renaming (.crdownload), is now recognized and uploaded as a PDF.
A file that really can't be read, like an unfinished download or an archive, now says exactly that instead of "content too large".
Study guides, cheat sheets, worksheets, and slides no longer fail when the PDF reader prints a note about the file while checking it; a document that rendered correctly now finishes.
A new chat that you reloaded before the first answer arrived now keeps working instead of refusing every message with "conversation not found".
A finished video, podcast, or document now opens reliably from its progress page instead of occasionally leaving you on the waiting screen.
The Create with AI row on your files now leads with Video and Podcast, the same order as the upload window; Flashcards moved further along the row.
Make next now leads with Video on podcasts and generated documents and with Podcast under a video lecture; Flashcards moved further along each row.
Canceling now has its own answer for "The results weren't good enough", separate from technical issues, and what you tell us about it goes to a person who looks into it.
Videos, podcasts, slides, summaries, and flowcharts stay on your selected chapter when custom instructions contain an unrelated title or description, unless you explicitly ask to use it.
The download window says exactly what you're downloading, previews page 1 of a PDF before you print or download it, and can email you a link back to the item for free.
Free video lectures show how much of a long document the short version covers, and choosing Too short in feedback offers a longer remake on Premium.
The end card on phones puts the longer-version offer, questions and replay in one place.
On a new account with nothing in it yet, Home now shows when you've reached a Free limit and how to get more, as it already did once you had content.
Chat answers about an uploaded Word or PowerPoint file right away instead of asking you to upload it again.
Summary PDFs keep Japanese, Greek and other characters that come from images or short identifiers, and print their study guide once.
Research PDFs show readable citations instead of raw citation markers.
Arabic text in generated PDFs now copies and searches as Arabic instead of unrelated characters.
Pages now typeset inline math: formulas inside a sentence, including those in pages made from a summary or research report, render as equations instead of LaTeX source. Type $…$ to add one, and click one to edit it.
Spreadsheet previews show FALSE and 0 formula results instead of blank cells.
AI File can now make a real Word document (.docx), with headings, tables and lists you can edit in Word. Pick Word Doc, or just ask for a Word file.
Flowcharts always include a decision diagram with labeled branches.
Quizzes no longer offer two equal numbers, like −0.0470 and −0.04700, as different answers, and questions name the ID, value or time they ask about.
Calendar files record when they were actually created.
Projects label Word and PowerPoint files as Word and PowerPoint, not PDF.
Signing out also signs out your other open tabs.
Muting a podcast keeps its video overview muted.
The task page only says it will email you when that creation's email setting is on.
Screen readers now hear the selected voice's name in voice pickers.

October 22, 2026

Scholarly in nine more languages

The app is now available in British English, German, French, Spanish (Spain), Spanish (Latin America), Brazilian Portuguese, Dutch, Korean and Simplified Chinese.
New accounts start in the language your browser uses; switch any time in Settings under Appearance.
Emails and Stripe checkout follow your app language.
AI language names in Settings now show in your app language.

Chats that recover, documents that still get built

When the AI provider stumbles mid-reply, the chat retries and can finish your turn on another model instead of stopping with an error.
Cheat sheets, study guides and other document downloads no longer fail when typesetting packages are slow to fetch.
Credits for failed generations come back faster.

From your file to a video lecture, faster

Uploading a file now leads with a narrated video lecture, then podcasts, chat and slides, and tells you when your first AI creation is free.
A file you drop on Home goes straight to making something instead of into chat.
Drop your notes on the cheat sheet, study guide or study schedule page and Scholarly starts making it as soon as your account is ready. There's also a new Study Schedule document built from your syllabus.
PDFs suggest a video lecture first in Make from this.
Record a meeting, office hours or a lecture and get a transcript, notes and a recap video.

Smaller changes

New chats start answering sooner: the first message creates the chat and starts the reply in one step. Stopping right after sending now keeps your question in the chat with Retry, the same as any other message.
PDF, Word and PowerPoint uploads are faster and more predictable: the workers that read your files now start already warmed up.
Uploading a PDF is faster, and Word and PowerPoint uploads are faster again: Scholarly starts reading the file while it is still being stored and converted.
Home, recordings, PDFs and videos open faster on a fresh visit: Scholarly asks for your lists and the document while the page is still loading, instead of after.
Chats stay smooth while a model shows its thinking: streamed reasoning repaints only the answer it belongs to.
Video lectures follow an uploaded storyboard more closely, keep each slide's title in place while points are added, and reveal each point as the narrator says it. When you pick Claude Sonnet or Opus, that model now designs every slide.
Follow-up messages in chats that used the code workspace, including PDF and file chats, start answering sooner: the workspace reconnects alongside the reply instead of before it.
Stripe receipts, invoices, the billing portal, Teams checkout and credit packs now follow your app language.
Cookie settings, the video lecture limit message, meeting allowance reset dates and the dates on podcasts, recordings, files and research reports now show in your app language.
The pricing link in an expense request email no longer points to a page that doesn't exist for some languages.
Going back to Home, Library or Projects returns you to where you were scrolled instead of the top.
Chats and folders open faster: Scholarly starts loading them as soon as you point at them. The progress page after Create starts loading while it opens.
Opening Scholarly in a new tab loads your account sooner, and long pages and chats respond faster while you click, type and read.
Recording on Android Chrome starts on the first tap again, instead of failing with "Microphone access was interrupted".
Long video lectures no longer lose a finished render when a final touch-up would run past the rendering machine's time limit; a render that won't fit now starts on a fresh machine.
Pressing Back from a finished creation goes back to where you started instead of to the finished progress page.
Safari keeps your selected files through checkout in more cases.
Topic requests like "the history of the video game industry" are no longer mistaken for a missing attachment.
Videos too short for interactive questions stop showing a long "taking longer than expected" wait.
"Try again with Google" on a failed sign-in always reopens Google sign-in.
Pricing now shows what one video lecture gets on each plan, with a sample to play, and Video lectures is in the top menu.
Prices outside the US show in your local format, such as 12 € or CA$42.
The plan screen shows how much of your free plan you've used, which payment methods work for you, and when you're charged in your own currency.
Downloading something on Free now offers a free view link first; downloads stay part of Premium.
If you leave checkout without finishing, Scholarly offers to pick up where you left off on your next visit, even on another device.
New pages for instructors and comparisons with Synthesia, Colossyan, Guidde and Pictory, plus exam results guides for PMP, CISSP, Series 7/SIE and CFP.
Marketing and help pages now state plan details accurately: downloads and printing are part of paid plans.
The changelog is now one entry per day, with the big launches first and smaller changes listed below.
If your browser uses a language Scholarly now supports and your account is still in English, Home offers to switch once.
On phones, the header Upgrade button uses a shorter label so page titles are no longer squeezed.
After you change the app language in Settings, Scholarly offers to make AI content in that language too.
Create tiles on the phone Home screen wrap long names onto two lines instead of cutting them off.
Long 1080p video lectures take the faster rendering path when a full render could not finish on one rendering machine, and a dropped connection no longer stops a render that was still running.
A finished video lecture is kept safe while it is touched up, so a rendering machine running out of time no longer costs you the video.
3D models built from a PDF finish loading in a background tab instead of waiting until you switch back to it.
Welcome, follow-up and subscriber emails now arrive in your app language in German, French, Spanish, Portuguese, Dutch, Korean, Chinese and Japanese.
A finished creation now always links its result in chat, even when the preview could not be loaded first.
Tool and comparison pages no longer show an error page when their text is slow to load.

October 21, 2026

A clearer progress page while you wait

A line at the top shows what Scholarly is doing right now, with an estimate of the time left once there is enough to go on.
A row of stages (reading, planning, narration, rendering, final check) shows how far along your creation is.
AI video lectures show their outline, tick off each section as it is narrated, and give a first look at the finished scenes.
You can start your next creation from the same source while this one finishes, and see what a longer lecture or more credits would take.
On phones, the page works as an app screen with Back in the top bar, the outline as a list, and a swipeable strip for the first look.

Smaller changes

Plan limits, upgrade buttons, and Premium labels now share one amber style across the app, so it is clear where an upgrade unlocks more.
Locked options in the welcome flow show a Premium crown instead of a gray padlock, and the credit pack you pick is highlighted.
Informational notes, such as "not verified as saved" or a deleted file, now use calmer colors so amber only marks an upgrade.

October 20, 2026

Summaries that cover every topic

Study Guides are now called Summaries and plan a section for every section and topic in your source.
Each Summary is reviewed once, and any section that is too thin or misses a topic is rewritten.
Every Summary ends with a study guide section: a review checklist, questions with model answers, and practice prompts.
A new Full Summary row on a PDF's Summary tab turns the instant summary into the full typeset Summary in one tap.
Summaries arrive faster, and the free instant summary on the upload screen is now labeled Quick summary.

AI Slides, infographics, and diagrams rebuilt

AI Slides uses the AI Video Lectures designs: real figures from your PDFs, diagrams, charts, tables, and equations.
Pick up to 30 slides on paid plans, choose a look or dark mode, and get speaker notes in the PowerPoint download.
Infographics use typeset panels from your sources, as a tall poster, 16:9 pages, or a painted Artistic poster.
Mind maps center your topic with labeled branch cards, and flowcharts show decisions as real Yes/No branches.
Story books keep the same characters and art style on every page, in square or wide 16:9 layouts.

Simpler project, PDF, and document pages

Projects open on one Add sources button and one Create menu with Video lecture, Summary, Flashcards and Podcast first.
Uploads, running generations and scheduled tasks sit together under In progress, and project contents are one list.
Sharing moved to a Share button in the project header with the link, invites, access requests and who has access.
Uploaded PDFs now have Document, Summary and Sources tabs, with a Make from this row on every tab.
Generated items like AI PDFs and slide decks open as one full-width page with a Refine bar and Make next.

Smaller changes

Finished meetings offer Turn this meeting into: SOP, Training doc, Summary, Quiz, Flashcards and Video lecture.
Meeting notes add a Key Facts section, and pressing Record now opens your microphone right away.
Transcribe and Deep Research on paid plans now show remaining weekly credits and ask to confirm large runs.
A transcription costs 1 credit, and a Deep Research run costs the credits of the model you picked.
Flashcard decks add cards for topics the first pass missed, and quiz and exam answer keys are checked against your source.
Free quizzes now draw from the whole deck, and the page-coverage meter sits at the top of the deck.
Infographics, mind maps, and flowcharts are checked for clipped or crowded panels before delivery.
A thumbs-down on the latest chat answer prompts a short question about what was wrong, without using your chat allowance.
Press ⌘K (Ctrl K on Windows) to open a command palette with recent items, quick actions, and grouped search results.
Settings now has six sections, and Plan & usage leads with your AI Creation Credits and Buy credits.
The Share window offers four choices: Only people you invite, Everyone at your team, Anyone with the link, and Public.
New links are not listed in Explore or search unless you choose Public; already public items stay listed.
People viewing a shared video or podcast without an account see one slim Made with Scholarly bar under the player.
Document Creator now applies customization profiles to the selected document format.
On phones, model choices, credit warnings, the podcast visual companion switch, and Use my script as written are in Settings.
Video resolution, logo removal, and Best Model options are visible in creation settings again.
Stopping a response before an answer arrives now offers Retry and keeps your question.
Chat edits stay editable at your chat limit and no longer shift off the screen on phones.
Fixed Create a quiz failing to open from a source page and chat errors when the app reloads between pages.
Chat replies that hit a temporary AI provider fault now retry or switch model instead of ending in an error.
PDFs now open on older Safari and Chrome versions, and load without locking the page on Safari 17.0 to 17.3.
Opening a finished video or cheat sheet from the waiting page is more reliable when the app reloads for a new version.
Finishing a meeting recording no longer reports a failure if your connection drops at the last moment.
Live meeting transcript lines now catch up faster when transcription slows down.

October 19, 2026

Notion and Dropbox in Chat

Connect a Notion workspace from Connections to search pages, read databases and comments, and use them as sources in Chat.
Connect Dropbox to search files, browse folders, and read supported documents in Chat.
You can connect more than one Notion workspace and reconnect expired access.
Both connectors are read-only, and you can disconnect either at any time.

Muse Image is the new default image model

Images in video lectures, chat, infographics and Image Studio now come from Muse Image, Meta's image model.
Muse Image produces clean, photographic results and can edit reference images.
Nano Banana 2 Lite, GPT Image 2.5 Flare and Sunburst, and Grok Imagine remain in the Image Studio model list.
Nano Banana 2.1 has been removed: saved selections switch to Nano Banana 2 Lite, and your existing images stay in your gallery.
For High quality, pick GPT Image 2.5 Flare or Sunburst.

Simpler Teams plans

A new Teams workspace now starts with your own seat, on your own plan.
Upgrade your team plan to add seats, invite teammates, and unlock Premium or Laureate for everyone.
Usage reporting, model controls, and shared brand kits are now part of the team plan.
The unpaid workspace is no longer called a trial, and Personal plans no longer show a Teams link.
The plan card in Team billing now lists what each seat includes.

Smaller changes

Podcast pages now open on a player card with chapters, speed and captions, and slides moved to their own Slides tab.
The podcast tab you last used now opens correctly after a reload.
Video lectures now open on the Summary tab, and Slides thumbnails play the lecture from that slide.
Standard video lectures can now add AI pictures where they help explain ideas, and you can still ask for no images or slides only.
Videos and podcasts use base credits up to 15 minutes, plus 1 credit over 15 minutes and 2 over 30 minutes.
New videos reserve the quoted credits and return unused duration credits if the result lands in a shorter tier.
Flashcard decks now have four tabs: Cards, Practice, Edit and Sources, and the Highlights tab has been removed.
About forty rarely used buttons and settings were removed or tucked away, so each page shows what you came to do.
The screen shown while a creation is made now has one progress ring and a typical time, and you can still leave at any time.
The header button now reads Upgrade Plan, and documents, videos, and podcasts no longer show the "Unlock the full experience" banner.
On the Free plan, AI Chat, flashcard decks, and video and podcast settings show fewer, shorter upgrade prompts.
Model menus label Auto by plan again: Auto (Free) and Auto (Premium).
New documents now open on the document itself, while your preferred reading tab, like Summary or Transcript, carries over.
Podcast lengths round to the nearest minute, one-page sources read "1 page", and First slide and Last slide are labeled correctly.
True/false practice questions now state the fact itself.
A file over your plan's size limit now names the limit and shows the plan that can take it, keeping your sources and settings.
The voice picker for videos, podcasts, and audio documents lists the exact voices that will speak, from a new set of 10 voices.
Narration in languages other than English is spoken by our multilingual voices.
The help center has new guides for 3D models, narration voices, and invoices and expensing.
The model guide now lists what every model costs in one table.

October 18, 2026

Image Studio rebuilt around your projects

Image Studio opens on your projects, and every image you make inside a project is saved to its files.
Each project's documents sit beside the image wall with image ideas from their pages; make one or several at once.
Click any image to edit it full screen: describe a change, drag over an area, use one-tap edits, or compare versions.
From a document reads a PDF or document and suggests covers, diagrams and social cards, citing the pages each came from.
Improve prompt (on by default) expands short prompts, and a brand kit of colors and a style note can be applied to any image.

Chat edits your story books and other PDFs

Ask chat to change any of your PDFs, including story books, slides, and guides Scholarly made, and it updates them in place.
Chat can redraw a picture inside a PDF, such as changing a character, while keeping the page's layout and art style.
Bigger edits run in the background, so you can keep chatting or close the tab until the updated PDF appears in the chat.
Redrawn pictures are swapped in place and the page text is left untouched.
Chat Auto now runs GPT 6 Luna on every plan.

New Gemini voices for narration and read-aloud

Video lecture narration, podcasts, read-aloud, and chat voice notes now use Gemini voices.
Read-aloud starts speaking in about half a second and applies the speed you pick exactly.
Castor, Naksh, and Atlas keep male voices, and the voice picker plays the new voices.

Smaller changes

Projects have a new Images button that opens Image Studio, and a prompt on the Projects page makes a quick image outside any project.
More like this makes a few alternatives from any image, and each edit uses one image from your allowance.
Image Studio has its own two-column layout on phones, with a prompt sheet and an edit screen.
Story book covers show the title once, with your name as the byline when you give one.
Chat now finds slide decks you made, and edits the open file or flashcard deck without asking which one.
Fixing a flashcard answer in chat refreshes its quiz questions, and chat can find your 3D models by name.
For podcasts, video lectures, and meeting notes, chat explains they can't be edited in place and offers what it can do instead.
Say "start a new chat" and chat opens one with your request already in the message box, ready to send.
Chat recognizes the story book, deck, or guide Scholarly made that you have open, so "the storybook you made" means that one.
Chat, PDFs, projects, and large folders load faster, and long conversations and audio transcripts stay responsive.
Playback no longer redraws the whole page, hidden tabs pause unnecessary polling, and failed PDF extraction is not retried on every open.
A meeting capture that never saved any audio now gives its meeting back, and meetings already lost this way have been returned.
Checkout lets you add a company name and VAT or tax ID to your invoice, and restates your plan and weekly AI Creation Credits.
The plan screen has an "Email your manager" link that drafts an expense request with your plan and price.
Signed-out visitors to shared pages, decks, PDFs, videos, and podcasts see an offer to make one from their own file.
Signed-out "copy to my account" links explain the waiting copy, let Back leave sign-up, and open a copied page once ready.
Score-release and results-day guides for the bar exam, MPRE, MCAT, USMLE, CPA, NCLEX and more now open with a video lecture module.
Several study tool pages add a "Turn it into a video lecture" link that opens the video lecture creator after sign-up.
Video lecture and podcast settings start on Free plan options; tapping a Premium option shows the plans without changing your selection.
New accounts see the video lecture first when choosing what to make from their material, on phones and desktop.
The Upgrade button shows a Free account's free creation: "1 free creation" until it is used, then "0 left".
After your free creation is used, Home offers its full version in a neutral row, and the automatic upgrade pop-up no longer appears.
AI documents such as study guides, slides, mind maps and infographics show a "Make the full version" row on the Free plan.
Free plan progress pages no longer show a credit-limit notice while a creation is still being made.
A Premium remake runs about 10 minutes for a lecture or 15 for a podcast; new ones can run up to 45 or 60 minutes.
Upgrade headings now name what was blocked, such as "Keep chatting", "Keep practicing", or "Download your files".
The upgrade screen describes Laureate by its 80 AI Creation Credits, Frontier models by name, and 1 GB files.
The line under the plans now names the professionals, researchers, and learners who use Scholarly.
Upgrading on a phone opens one sheet with the Premium price, a Yearly/Monthly switch, key features, and one Get Premium button.

October 17, 2026

Home keeps Chat in the box, tools one click away

On desktop, the box on Home is now just Chat.
Documents, flashcards, research, pages, 3D models, study guides, slides and more sit below it as buttons that open each tool directly.
Page now opens a blank page straight away, like starting a note.
Phones are unchanged.

Video lectures recover instead of stalling

If a video lecture's rendering environment disappears partway through, Scholarly now moves the run onto a fresh one right away.
Lectures no longer sit unfinished for hours.
If it can't recover, the lecture fails and returns your AI Creation Credit instead of leaving you waiting.

A middle-sized credit pack

You can now buy 35 AI Creation Credits for $30, between the 10- and 120-credit packs.
Available for paid personal plans and team members.
Purchased credits never expire.

Smaller changes

Images in pages created or updated through chat now stay visible as illustrations instead of turning into text links.
Chat can now read saved pages when preparing PDFs, with clearer guidance on image allowances and reusing existing illustrations.
DeepSeek V4.1 Flash now uses the October 17 model update in Chat and AI creation tools, at the same credit rate.
Finishing signup now creates your starter pages once instead of twice, and the assistant opens the way you left it.

October 16, 2026

3D models (Beta) built by AI from a PDF or description

Create a 3D model from Home or the New menu and describe a process, structure, system, or molecule in the chat beside it.
Start empty or from a PDF in your library, and the model is built from what its pages say.
The AI writes the model as real three.js with labels and animation, checks its own render, and gives you a draft to preview and apply.
Small chat edits use no credit, and a full build costs 1 AI Creation Credit after you confirm, returned if the build doesn't finish.
On a paid plan, export a 4K image, a turntable video, or a .glb file for PowerPoint and Keynote.

Manage more of your workspace through chat

Ask chat to unpin all your chats and content, rename or restore chats, and change workspace preferences.
Chat can also update project instructions and manage background tasks.
Preview sharing changes and confirm the exact audience before access changes or invitations are sent.
Library listings now continue beyond the first page.
Chat only soft-deletes content, and permanent deletion stays in your manual controls.

Calmer flashcard study and quiz questions

Flashcard decks open in one step with nothing shifting, and the card itself flips with a click, Space, or tap.
On phones, decks have a simple ‹ Back · title · ⋯ bar with one row to step through cards.
A deck's Questions tab makes questions once in the background, shows them as they're ready, and keeps them when you return.
Quizzes get a full-width Next on phones, and question counts no longer double up.

Smaller changes

Rename chats, pages, PDFs, decks, and other items by clicking the title itself; the pencil button is gone.
A chat's pin now sits under its icon menu, and the sidebar's Chat link replaces the chat panel's full-screen button.
New quizzes are named as quizzes, and projects have a single Add menu that starts with From library.
The phone welcome flow now takes a link and keeps your chosen creation when you skip.
Log out lands straight on the sign-in page, and the System theme follows your device when it switches between light and dark.
A source keeps one name everywhere (“paper”, not “paper.pdf”), Library search results match the Library, and pinning shows at once.
Progress pages are titled by your source, show the model's name instead of its ID, and offer an upgrade when you're out of credits.
A deleted or unshared item now opens inside the app with a way back to your Library.
Library search previews no longer show a chat's internal working notes.
When chat refuses a model or allowance your plan doesn't include, Scholarly opens the upgrade options for that limit.
A recording stopped by a plan limit shows its upgrade options without a second warning on top.
Video summary, transcript, and question tabs, and creation window source tabs, keep scroll, search, and choices when you switch.
Headings, body text, and what you type now show in full white instead of muted gray in dark mode.
AI creation and source panels show one spinner while loading, and answer fields, search boxes, and upgrade buttons match the app.
After you press Create on an upload, you go straight to the progress page, which shows each step and opens the result when ready.
When a plan limit stops one creation in a welcome or post-purchase project, Scholarly shows how to upgrade while others continue.
Starting a quiz now goes straight to its progress page without briefly showing the page behind it.
Your workspace opens with less to download, and the welcome screen no longer waits for an extra download after sign-up.
Text typed into Sign up or Log in while the page is still loading is no longer cleared.
Sign up now says when an email address can't be used for an account, instead of showing a general error.
After picking your first creation and pasting text or a link, Next goes straight to that creation's settings.
Scheduled chat tasks on Home match other scheduled creations, and canceling or starting one early asks a clear question first.
The sidebar keeps hover and the selected page visually distinct.
Plans, credit packs, file previews, and PDFs open with a single spinner, and phones keep a Back button while a step loads.
A recording with damaged audio offers to upload a working file, and the meetings allowance in Billing matches your other allowances.
An app update no longer shows a 500 error code.
Flashcards, quizzes, podcasts, and video overviews from a long YouTube video now wait for its transcript to finish.
Word previews show common legacy bullet symbols, and calendar previews show commas, semicolons, and line breaks as written.
Word, PDF, and slide files made in chat name their sources in plain text instead of unreadable codes.
Autocomplete no longer inserts stray markup, and returning from checkout takes you back to what you were doing.
When chat adds images to Image Studio, it shows one short line that opens the Studio and tells you if an image failed.

October 15, 2026

Schedule background tasks in chat

Ask chat to start a background task at a future time.
Find your scheduled work on Home.
Start a scheduled task early or cancel it from chat.
Ask for a completion email for scheduled or immediate tasks.

Faster loading on phones and slow connections

Home and public tools load less code before you can use them.
PDF and diagram tools load when needed, and PDFs start loading when you select them.
The phone navigation menu responds to your first tap while the rest of the page is still loading.
If projects fail to load, the phone Library and Home show a retry button, and loaded projects stay available during a refresh.
Calculator links work from the first tap, and AP score calculators keep your first edits while the page loads.

Smaller changes

Diagrams appear without toggling the preview and stay readable with Reduce Motion enabled.
Recordings with audio that cannot be decoded stop retrying and explain when a working file or a new recording is needed.
Fixed saves getting stuck when an older and a newer browser tab sync the same edit, without overwriting newer work.
Fixed pages with a divider failing to open, and fixed voice chat errors when displaying tool results.
Fixed repeated chat failures with image attachments between 10 and 32 MB, including when continuing a conversation.
If part of a page fails to load, a Reload button brings you back to the same page.
Creating from the Chat limit screen keeps the document you are viewing selected.
Team libraries show the original names of Word and PowerPoint files.
When you question an exact value, chat checks the source page again.
Transcript and slide controls have clearer screen-reader labels.
The PowerPoint download prompt now describes slides instead of a video.
On the Free plan, Billing shows how many meetings you have used.
Video questions show every passage they cite, matched to the right moment in the video.
A delayed Quiz upgrade prompt no longer appears after you switch to Exam mode.
Personalized emails with findings from your sources arrive at most once a week, and less often if you do not reply.
Help search results and article lists now show each article's own summary and update date.

October 14, 2026

Reply to messages, with menus and shortcuts

Reply to an earlier message, an answer block, or selected text, then jump to the original and back to your reply.
Right-click a message to reply, copy, listen, edit your own message, or retry an answer.
Press R to reply, L to listen, and Alt or Option with the Up and Down arrow keys to move between messages.
On phones, swipe right on a message to reply, or long-press it for its actions.
Retrying an answer on a phone asks you to confirm first, since it uses credits.

Image Studio in your sidebar

Image Studio is now a sidebar destination on desktop and mobile, replacing the image tool on Home.
Image Studio conversations share your chat history, pins, voice notes, and background tasks.
Images use a separate allowance: 2 lifetime images on Free, 40 per week on Premium, and 80 per week on Laureate.
These images no longer use AI Creation Credits; image creation is in beta and allowances may change.

Credit cost and allowance changes

Claude Sonnet now uses 2 AI Creation Credits instead of 3, and Best Model uses its selected model's cost (currently Sonnet).
Opus remains at 3 credits, and videos and podcasts longer than 15 minutes still add 1 credit.
Extended research and Go deeper now use 2 credits (Standard research still uses 1), and a failed run returns both credits.
With only 1 credit left, Extended research opens your plan limit and does not spend that credit.
Free includes 3 meetings total, up to 60 minutes each, with no reset; Premium and Laureate keep their weekly allowances.

Smaller changes

Public pages and Help articles open faster, with less to download.
Ask chat to rename itself, change its icon, or pin and unpin the thread; your chat panel and draft follow you into Image Studio.
Model menus are simpler, showing model or Auto tier names without reasoning-effort or Thinking-status details.
Each AI creation, including scheduled work, has an “Email me when it’s ready” toggle that starts on.
Home and checkout recovery now separate unlimited uploads and AI Chat from weekly AI Creation Credits, which research also uses.
When creating a video, choose Help me understand, Help me apply it, or Refresh my memory, and set a starting point or focus.
Large image attachments in chat are resized for the AI when needed, so you no longer need to compress them first.
Chat recovers better from temporary connection failures, code workspace setup failures, and AI provider changes.
Incomplete podcast and workspace organization requests can now be corrected in the conversation instead of stopping the turn.
Chat and suggested questions keep working when your browser cannot provide a timezone.
Reload on the app-update error screen now retries the failed file and reopens your page in the same position.
Flashcard modes switch to a compact menu when the sidebar or chat leaves too little room for the tab strip.
Research remembers your Produce, Evidence, Depth, and Extended choices in this browser for your next session.
Voice conversations now use Grok Voice 3.0.
Fixed videos, podcasts, recordings, and file previews wrongly showing "can't play this", especially in Safari and on iPhone.
Uploads and creations that time out now say so instead of showing an internal error message.
Flashcard sets, page drafts, and other chat results now use the same compact card as tasks and files.
On a phone, each chat in your history has a "..." button with Rename, Pin, and Delete; before, a phone could only pin.
Answers on a phone are set slightly smaller and tighter, so more fits on the screen.
Podcast previews in chat use the same scrubber as voice notes, and cards, quotes, and pictures share one corner radius.
Fixed citation badges sitting inside empty brackets, often split across two lines.
Answer headings are larger, chat card buttons are black or white instead of blue, and image generation shows a small spinner.
Chat history has smaller section labels and a larger pin button on phones, and its search no longer zooms the page on iPhones.
Shared chat pages show your messages in the same solid bubble as the app.
Emoji and GIF reactions left on answers in the past are no longer shown.
Message hover actions no longer cover the message below, and chat action buttons stay fully visible above the composer.
The Sources button now includes saved web results even when an answer uses ordinary links instead of citation badges.
Drafts show in full when a chat opens, and the scroll-to-bottom button appears above the message box only after you scroll up.
Pinned chats show a plain filled pin, code blocks are readable in light mode, and the send button is solid white in dark mode.
Answers are plain text on the page again instead of a gray bubble per paragraph, with Sources next to Copy, Listen, and Retry.
Editing a sent message now ends with a small Cancel and a compact Save & send button, and the "Created Memories" label is gone.
Swipe left with two fingers on a trackpad, or drag left on a phone, to see when each message was sent.

October 13, 2026

A lighter, messaging-style chat

Your messages are solid bubbles on the right, and each paragraph of an answer is its own bubble on the left.
While chat works, one quiet line says what it is doing, then what it did, such as "Thought for 12s".
Click that line to see every step and the thinking behind the answer.
Copy, listen, retry, rating and sources appear on hover or tap, and stay in view on the latest answer.
Small centered timestamps mark where a chat starts, pauses of 15 minutes or more, and each new day.

Chat memory across chats

Chat's memory is now a set of named files you can read, edit and delete in Settings → Chat.
Ask about something from another chat, and Scholarly finds that chat and pulls up what was said.
In long chats, chat can look back over its own earlier messages instead of asking you again.
A project's memory is now one note you can edit freely in its Context tab.
Deleting a chat removes its notes from memory, and chat tells you when a chat you mention was deleted.

Video lectures and podcasts: new models and clearer costs

Paid plans start on a 15 minute length, and anything longer than 15 minutes uses one extra AI Creation Credit.
The Create button now shows how many AI Creation Credits a video lecture or podcast will use.
Gemini 4 Argon and Muse Wonder are now available for video lectures and podcasts, at 4 AI Creation Credits each.
Claude Opus 5.5 now costs 3 AI Creation Credits instead of 4.
Standard video lectures use one continuous script, and Claude Sonnet or Claude Opus plans the teaching and fixes slides.

Smaller changes

Messages you send within two minutes of each other now sit together as one group.
Tasks, creations, pages, decks and files in chat share one compact card, and long code opens with "Show all lines".
On a phone, wide tables in chat scroll sideways under a pinned first column.
New chats get a fitting name and icon after the first answer, and are renamed if the subject changes, unless you named them.
Chat icons come in ten shapes and eight colors and show in the sidebar, Home, the chat page and chat history.
Right-click a chat and choose "Edit name and icon", or edit both in the chat panel beside your sources.
Rename a chat or change its icon by asking in the chat, or on a phone by tapping its name at the top.
The length you pick for a video lecture or podcast is now the most it will run, and Custom opens the slider.
When a team's paid plan ends, the team Overview now says so and explains that shared work, projects and members are kept.
On a phone, plan options now show the line under each feature that says what it covers and what Free allows.
Canceling a monthly Premium plan because of price now offers 15% off your next 3 months instead of a switch to yearly.
View-only research links now hide Export, Create page and file downloads, including browser downloads on audio and video.
View-only links now hide Download on podcasts and recordings, and Save to your account on pages and decks.
Invited people who have not confirmed their email are now told so and sent the verification link.
Folders, projects, flashcard edits, long chats and streaming now download and send less data.
Duplicating a PDF no longer downloads and uploads the same file again.
Home, panels, file viewers, public writing, math and study tools, and blog pages load faster.
Text files download once for Preview and Edit, and recovered drafts stay usable if the original file fails to load.
Fixed a finished task flipping back to an older status.
Sidebar folders and chat drafts open faster after the first time, and switching accounts stops any reply still running.
Audio keeps playing while you browse Creations, and your transcript search is still there when you return.
Deep Research flags a source over your plan's limit right away, and Laureate and Teams plans can combine up to 20 sources.
Research now explains what is blocking Start, and Export appears only once a report is ready.
With no uploads left, the upload window opens on your Library and points to Recently deleted when it is empty.
Dropping a file on the Library page now starts an upload, and unsupported files are explained in the upload window.
Fixed library search in create windows showing "No results" next to "Load more".
On a phone, empty projects start with Add sources, each project row has a menu, and failed uploads list under "Not uploaded".
Adding a source from the team library's Add button now shares it with your team.
Addresses added during team setup that need a paid seat are now kept and invited when you choose a plan.
Join links opened during team checkout now say the team is finishing setup, and survive going back from sign-up or log in.
Trying to create a second team now offers to open your current one, and invite sign-up shows the invited address.
Team members who open plan options now start on the Personal tab, where they can make a purchase.
Canceling a creation no longer also shows a "We lost track of this generation" notice.
Fixed Safari discarding an earlier saved upload selection when a new one could not be saved before checkout.

October 12, 2026

Chat can run tasks while you keep chatting

Chat can hand bigger jobs, like a calendar file or a spreadsheet from your sources, to a task while you keep chatting.
A card in the thread shows what the task is doing and lets you cancel it.
The result is posted in the same thread with the file attached, even if you closed the tab.
Unread results show a dot in the sidebar and chat lists, a row on Home, and a browser notification if allowed.
For long tasks, the card can email you when it is done, sent only if you have not opened the result by then.

Video lectures play through and end with a next step

Questions appear over the lecture with Answer and a close button while it keeps playing, and leave on their own if ignored.
Answering holds the lecture while you read the explanation, and Continue or closing the question resumes where you were.
When a lecture ends, the player offers to answer its questions, replay it, or remake the free short version at full length.
A question near the end no longer leaves the lecture paused one second before its last frame.

Shared folders open without an account

Folders set to Anyone with the link now open for people who are not signed in, instead of sending them to log in.
Visitors see the folder and the names of its items, and each item opens if it is shared too.
A password, expiry date, and Revoke on the folder's link apply to everyone who opens it.
The owner's email address is not shown to signed-out visitors.

Smaller changes

A chat open in two tabs or browsers stays in sync: messages and streaming answers appear in both at once.
On phones, lecture and podcast questions open only from Answer, tapping the card no longer pauses playback, and buttons are larger.
Flashcards on phones have a full-width Next card button and a hint that you can swipe to change cards.
The video fullscreen button is no longer cut off on narrow phones, and on Android fullscreen asks to turn to landscape.
The creation progress page now says you will be emailed when it is ready, shown right under the status on phones.
On phones, the PDF page now remembers the last tab you used, as it already did on desktop.
Create windows from Home no longer pick an empty project for you, and the Library tab's Library link shows everything.
Create with AI on a spreadsheet in your library now opens the creator instead of saying the source cannot be used.
Research you opened but never started no longer sits in Recent as "Preparing research…".
At the free chat message limit, the Send arrow on Home and buttons like Discuss now open the plans instead of doing nothing.
A question typed on Home now starts a new chat instead of joining the last chat you had open.
On phones, Return in the chat box adds a new line and the arrow sends.
The chat box now says to remove or retry a failed attachment, and keeps a Home question if you leave to upgrade at the limit.
Share stats now count views of public links by people who are not signed in.
Signing up from a shared page's header returns you to that page, and on phones the sign-up prompt is a short bottom sheet.
Revoke link now asks once and says that links already sent stop working for good.
On phones, the Share button sends the same link the Share window copies, including password links, and opens Share when revoked.
The Review & approve email opens the shared chat where you can approve; signed out, the download button says Create a free account.
The sign-up prompt on a shared page, deck, PDF, or video saves a copy to your new account; on other items it no longer says Save.
Share invite emails no longer call everything a page, say which address to open it with, and note when someone already has access.
On a shared page you do not own, the download window offers Copy link instead of the owner's sharing options.
The button in the email sent a week after canceling now opens Billing with your discount, not a missing page.
After canceling, Settings stays on Billing so you can see your plan's end date and turn renewal back on.
The pause offer while canceling now says the pause starts at once and your account uses Free plan limits while paused.
Choosing "missing features" as a cancel reason now says the feature name is needed to continue.
Team members who pay for their own plan now see it in Settings → Billing beside the team's, with Pause and Cancel.
Pressing Start a team while signed out, then logging in to an existing account, now opens team setup instead of Home.
Team setup lists addresses that could not be invited with the reason, and resending an invite confirms after it is sent.
On phones, Answer questions at a lecture's end brings the Questions tab just under the video, and reopened questions scroll into view.
A private page opened while signed out now says to sign in if it is yours, and Sign in keeps your link.
Settings links in our emails now ask you to sign in first, then open the tab the email pointed to.
Video lecture diagram slides size boxes to their text, so boxes no longer overlap, and long labels no longer break mid-word.
Figures cut from your PDF for video lectures keep their labels whole, leave out the page heading, and now fill the slide.
Slide decks saved as PDF work better as sources, and PDF paragraphs and lists become slide text instead of page pictures.
Step-by-step slides connect their steps with a line, and long figures like “280 million” stay on one line.
A new AI Models & Defaults help guide lists defaults and reasoning settings for Chat and AI creations.
When video lectures and podcasts need a paid plan, you are told when choosing them, with free creations one press away.
On one-source plans the file picker takes one file, phones show upload limits first, and Create says when it is reading your PDF.
Signing up from about 35 more tool pages opens the promised creation; study schedule and calendar pages open chat with your request.
Enter on the forgot-password page now sends the email, and account-exists notices on login and sign-up scroll into view on phones.
With your free upload used, Home's Upload tile opens your Library, where that file is.
If a renewal payment fails, Home now says so with a button to update your card, then returns you to Billing.
The checkout button stays busy until the payment page opens, and a second press no longer expires the loading page.
With a paused plan, Resume and upgrade buttons now open Billing to resume the plan instead of a refused checkout.
For Free accounts, the download window's button now says it upgrades instead of Download.
Returning subscribers get their plan activated on the welcome screen, and files picked before paying come back on Upload.
The podcast creator now explains when podcasts need a paid plan.
Opening a chat Scholarly prepared from your material now clears its Home card and unread dot.
On smaller laptop windows, pages with the chat panel open no longer shift as the panel narrows while loading.
Podcast questions now arrive closed without jumping the page, and a finished episode offers its questions and a replay.
The last flashcard says Finish, swiping past it finishes the deck, and starring on the finished screen no longer closes it.
Escape now closes a question opened on a flashcard.
On Free, skipping a quiz question no longer uses a free answer, and exams have a Submit button after the last question.
When free quiz answers run out, the practice exam included with your plan is offered beside the upgrade.
Recording notes open on their own when ready, and reaching the time limit shows one message instead of one every second.
Suggested next creations start from the same source, and a finished creation's notice stays up with an Open button.
A YouTube source page no longer stops responding if switching tabs or a plan limit pauses the video before the player loads.
Instructions about word order in Arabic or another language no longer change your creation's selected language.
Chat no longer denies reading a source it already read, so follow-up questions build on earlier grounded answers.
Exam feedback now follows your graded answers and the source instead of contradicting a correct score.
On medium-width windows, the flashcard assistant sits beside the card instead of covering it and the Next button.
Video lectures treat text typed with your sources as instructions and say requested codes, status words, and non-English terms exactly.
Chat's history and source buttons have screen-reader names, and the welcome tour keeps keyboard focus inside the dialog.
Teams pricing now shows 15 AI Creation Credits per member each week everywhere.
Help articles now say Free podcasts and videos target about two minutes.
Asking chat for flashcards with a PDF or other source open now starts a deck in Background Tasks again.
On phones, log in, sign-up, and password reset forms use the full width, and pages and app screens no longer stack extra margins.
Deck Complete now leads with a practice exam and questions on those cards, showing what your plan includes.
When free quiz questions run out, the screen leads with your score and missed questions, and says how many more are ready.
On Free, a deck made from part of a long PDF now says how many of the document's pages it covers.
Quiz limit and Deck Complete screens now use the full width of a phone.
Studying a deck no longer shows the Edit with AI and Looks good prompt; Refine with AI stays in the deck editor.

October 11, 2026

Long video lectures finish much sooner

Standard video lectures render several times faster.
The render step of a 45-minute lecture took 20 to 35 minutes and now takes about 5 in our tests.
Picture and sound are unchanged: the same resolution, frame rate, and loudness.
Narration begins about half a minute sooner because a lecture's workspace starts while its outline is being written.

Warnings for unreadable sources and clips with no speech

The creation window warns when a library source is a sign-in page, mostly menus, unreadable, or empty, and suggests what to upload.
You can still go ahead, and freshly uploaded files with the same problem get the note on the progress page.
A video, recording, or clip with only music or applause says it has no spoken audio when picked, and no credit is spent.
A meeting recording that picked up no speech is reported as such, and its audio is kept so you can play it back.
Library audio with only music or applause is marked as having no speech, and on the Free plan the upload is returned.

A calmer Customize step when you create with AI

The Customize step for videos, flashcards, podcasts, slides, study guides, documents, and other AI creations is shorter and quieter.
Styles and presets are one row of names with a one-line description, and settings are a plain list of labels and controls.
Schedule, Plan with Chat, and other rarely changed options moved into a More options card, so the footer is Back and the create button.
Choosing a document type shows every type by name on one screen, with its description appearing as you point at it.

Smaller changes

Finishing onboarding now saves reliably, so closing the tab soon after no longer sends you through onboarding again.
Images attached in AI Chat now show as the picture in your message and as a thumbnail while you write; several appear as a grid.
Selecting an attachment in the chat panel beside a file, PDF, note, recording, or other source now opens its preview.
Leaving checkout before paying keeps a dismissible Finish upgrading row on Home for seven days that reopens checkout in one press.
After a subscription ends, Home keeps a renew option for 30 days; AI Creation Credits bought while subscribed stay usable on the Free plan.
When free AI Chat messages run out, chat offers your free AI Creation Credit for a video lecture, study guide, or flashcards.
Finished meeting recordings lead with Make from this meeting, and Free plans see how many meetings are left this week.
The flashcard creator shows which pages of a long PDF your plan covers, and the video length setting shows each plan's length.
Next-step suggestions on the progress page now match the source, such as a recorded lecture, a contract, or a slide deck.
Podcast, video lecture, and infographic instructions are read in any language and can set length, speakers, or orientation.
Canceling reasons are understood in any language, so the options offered next match what you wrote.
Shared items are checked before appearing on Explore; spam, promotional, and near-empty items stay unlisted but share links still work.

October 10, 2026

Chat now starts on Claude Haiku 5.5

Auto in AI Chat now uses Claude Haiku 5.5 on every plan, Free included.
On paid plans, the Thinking switch now works on Auto: leave it on for harder questions, or off for the fastest replies.
GPT 6 Luna is still in the model menu, and the creation tools keep their current models.
AI Chat now recommends GPT 6 Luna, Gemini 3.8 Flash, and Claude Haiku 5.5.
Mistral Large 3 and Muse Spark 1.3 remain available under More Models.

Delete chats and browse history beside the chat

You can now delete a chat, and Undo in the confirmation brings it back.
A deleted chat leaves your chat history, the sidebar, Home, and search right away, and its shared link stops working.
On desktop, Delete chat is in a chat's ⋯ and right-click menus, and sidebar chat rows show ⋯ on hover.
On desktop, the clock button in full-screen chat opens a right panel with search, pinned chats, and recent chats.
The panel stays open as you switch chats, marks the open one, and each chat has a ⋯ menu to pin or unpin it.

See your weekly credits before you run out

On paid plans, every create window shows this week's remaining AI Creation Credits, with Add credits at three or fewer.
The weekly limit notice names the day credits return and shows the smallest pack's price, one click from checkout.
After buying a pack, you return to your page and your creation reopens, including files added in the Upload window.
Premium subscribers at the weekly limit get a simpler screen that sets a one-time pack beside Laureate.
The credit pack window now has one selected pack and one Buy button on every screen size.

Smaller changes

Videos, podcasts, research, and AI-written documents now start within a minute of pressing Create, not about two minutes.
Video lecture and document generations now continue from the sources they have if their workspace is replaced mid-run.
Failed videos, decks, podcasts, and documents now stay under Didn't finish on Home with the reason and a way to retry or fix.
Dismissing a failed creation removes it on all your devices, and Home shows the newest three with a way to dismiss them all.
Free plan: if an uploaded file can't be read, your free upload is returned and the file moves to Deleted.
A meeting recording with no speech no longer counts toward your weekly meetings.
Podcast episodes now open as soon as the audio is finished, with questions, producer notes, and related resources appearing moments later.
Opening a podcast the moment it finishes now shows the finished episode, not a leftover “queued” screen.
Short podcasts delivered far shorter than requested are now regenerated, and your credit is returned if that fails.
Right-clicking a chat in the history or on the chat page's pinned and recent cards now opens its menu and keeps it open.
If your plan reads 32 pages per document, you can now pick any 32 in a row from a PDF, such as pages 90 to 121.
Free plan: adding cards from a PDF in the flashcard editor now reads as many pages as your plan covers, with no upgrade prompt.
Documents, text files, and images up to 50 MB are read when chosen, so files moved by iCloud, Drive, or other apps still upload.
A file that can no longer be read now offers Choose file again instead of a Retry that could not work.
The "taking longer than usual" notice now appears only when a podcast, slide deck, study guide, or video is really running late.
Finish and failure notices now stay longer with a button to go to the result, and finished documents no longer leave a spinner on Home.
Study guides, slides, and other documents no longer sit for five minutes when the first read of your sources hangs.
YouTube links we could not check are no longer reported as private or deleted, and links with a playlist import the linked video.
Flashcard decks start generating right away and open about twenty seconds sooner, with practice questions finishing in the background.
Sources that only name topics, such as a title or topic list, now get a flashcard deck on those topics instead of an error.
Phone photos of notes are now read the right way up, so handwriting shot in portrait is no longer rejected as unreadable.
Try again on a failed flashcard deck now shows live progress right away instead of the old error.
Older chats started from a page, PDF, or project can be shared again.
Follow-up research requested in chat on a research report now appears on the report right away, without a reload.
In voice mode, talking over the assistant after opening another source keeps your earlier question tied to its original source.
Plan with chat from a creation window opened on a chat now always sends your planning message from a clean message box.
The Talk button is back in the Home chat box on desktop and phone, starting a live voice conversation in AI Chat.

October 9, 2026

One chat that follows you across Scholarly

Chat is now universal across Home, Chat, projects, and your content, keeping your thread, history, and draft as you navigate.
Older per-item chats now open in global history, and you can start a new thread whenever you want.
Each message remembers the page, PDF, video, or project you had open, and long threads load older messages as you scroll.
Answers start sooner and long threads stay quick, since older turns are summarized as a thread grows.
Pinned chats lead on Home, the sidebar, the Chat page, and the chat panel; with none pinned, your most recent chats show.

One Filter button in your library

Library and project toolbars are now a search field and a single Filter button.
The Filter panel holds the view (all content, pinned, shared, or deleted), type, grouping, and sort order.
The Filter button shows how many filters are narrowing the list.
Opening the library shows one loading indicator instead of two.

Video lectures that are easier to follow and wait for

Narration is written for listening: one idea per sentence, with full lists and tables kept on screen while the voice explains them.
The Length setting shows how long a lecture usually takes to make, such as about 55 to 95 minutes for a 45-minute lecture.
The progress page's expected wait now scales with the length you chose, so long lectures are not flagged as slow too early.
A Math video lecture is now assembled from the scenes that drew, instead of failing completely when one animation won't draw.
Transcripts and captions show readable math like a/b, θ, aₙ and √x instead of raw TeX markup, and narration says formulas aloud.

Smaller changes

New one-time AI Creation Credit packs: 10 for $10, 120 for $90, 500 for $225, or 2,200 for $999 USD, excluding tax.
Team Owners and Admins can top up a member's credits, existing credits never expire, and the phone picker uses one bottom sheet.
Nano Banana 2.1 replaces Nano Banana 2 in Image Studio; saved selections carry over, and Nano Banana 2 Lite remains the default.
Files too large for your plan now show their size beside your limit and name the plan that accepts them, on every upload screen.
Chat no longer prints every library item it finds into the conversation; the lookup is folded into the steps above the answer.
Fixed Word files, PDFs, and finished Slides sometimes showing a blank page and "Failed to load PDF" in the viewer.
"Made from" on Slides, Podcasts, and Flashcards now shows your source's real pages and fills in as soon as a creation finishes.
Video lecture transcripts keep times and codes exactly as written, such as 18:07:45 and PRX-73C2.
Arabic and Hebrew text in generated Slides and documents is read in the right order, so identifiers are no longer reversed.
Fixed the podcast transcript on phones cutting off speaker names and the first letters of each line.
Free podcast and video lengths are described as about 2 minutes everywhere, and finished free videos are labeled as the short version.
Renaming, deleting, or reporting an audio gap in a meeting now works while the meeting is recording or writing its notes.
Fixed a failed meeting delete leaving the meeting in your library without returning its AI Creation Credit.

October 8, 2026

Voice notes in chat

Ask chat for a voice note, such as "send me a voice note on this" or "say that out loud", and it replies with a short spoken message.
Voice notes play right in the conversation, in your chosen assistant voice and your language.
Tap play to listen, scrub to any point, or open the transcript to read along.
Chat sends voice notes only when you ask for one, and they also appear in the chat's files so you can download them.

Claude Haiku 5.5 replaces Claude Haiku 4.5 in chat

Claude Haiku 5.5, Anthropic's newest fast model, now powers the Claude Haiku option in chat on every plan.
It follows instructions more closely, reasons when a question needs it, and can write much longer answers.
Conversations that had Claude Haiku 4.5 selected switch to Claude Haiku 5.5 automatically.

Video lectures hit their length in more languages

Video lectures with the narrator set to Auto are now planned at the pace narrators have been measured reading your language.
Languages that read more slowly, such as Arabic, Dutch, and Danish, get a script sized for your chosen length from the start.
A language chosen by its own name, such as 日本語 or العربية, now keeps its own measured pace.
Hindi, Nepali, Bengali, Dzongkha, and Thai lectures now plan their narration against a correct word count.

Smaller changes

Email sign-in links now work for accounts that normally sign in with Google or Apple.
Wide tables and figures from your PDF now span the full width of video lecture slides, so every row stays readable.
Video lecture diagram arrows now draw only after both boxes appear, and stop at the box edge so you can see their direction.
Fixed boxes overlapping in left-to-right video lecture flow diagrams with four or more steps.
Video lecture diagrams and comparison slides now fill in while the narration explains them instead of near the end.
Video lecture charts no longer label unnamed data series "Series" in English in non-English lectures.
Video lectures finish sooner: the AI now redraws only the slides it fixes, and slow-loading slides retry much faster.
Video lectures of 20 minutes or more now get up to six generated illustrations instead of three, and repeat pictures less.
Fixed some video lectures failing at the last step, most often ones narrated from your own script.
Video lecture chapter markers now jump to the exact scene where each topic begins.

October 7, 2026

Six new animated scene types in video lectures

Function graphs in standard video lectures draw themselves and can shade an area, show a tangent, and mark key points.
Tables fill in row by row and highlight the answer, and 2×2 frameworks build cell by cell and focus the recommended quadrant.
Funnels narrow stage by stage with their figures, and cycles loop around with an arrow that keeps traveling.
Code walkthroughs highlight lines step by step with a note for each.
Lectures that need these visuals no longer fall back to a slower way of building the video.

October 6, 2026

Video lectures planned by GPT 6 Luna

Every video lecture's script and scene plan are now written by GPT 6 Luna (high reasoning on paid plans).
The model you pick builds the video from that plan.
Long lectures made with Claude models no longer come out far shorter than the length you chose.
Video lectures from long PDFs start building sooner because their pages are read in parallel.
Claude Opus 5.5, Claude Sonnet 5.5, GPT 6 Astra, and GPT 6.1 Sol videos now finish on GPT 6 Luna at high reasoning after a model failure.

Credit changes for Claude Opus 5.5 and GPT 6.1 Sol

Claude Opus 5.5 now uses 4 AI Creation Credits per creation, up from 3.
GPT 6.1 Sol now uses 3 AI Creation Credits per creation, the same as Claude Sonnet 5.5.
GPT 6 Luna and Auto still use 1 AI Creation Credit per creation.

Smaller changes

Images in Illustrated, Freestyle, Short, and Standard video lectures now hold still and change with fades instead of zooming.

October 5, 2026

Meetings no longer use AI Creation Credits

Meetings no longer use an AI Creation Credit, and the transcript and notes are included.
Free includes 3 meetings a week, each up to 60 minutes.
Premium includes 20 meetings a week and Laureate includes 100, each up to 10 hours.
The allowance resets Mondays at 00:00 UTC, and if it is full, Scholarly says so before recording and tells you when it resets.
If a meeting fails, you can retry its notes or replace the audio without using another meeting or a credit.

Smaller changes

Making a podcast or video lecture from a video with no spoken audio now says so instead of waiting for a transcript.
PDFs now open reliably when storage briefly drops a connection.
If a video lecture still will not play after you retry, you can open the video file directly.
Signing in to another account in a different tab now tells the old tab to reload instead of retrying saves in the background.

October 3, 2026

Narration in your language for 30+ more languages

Video lectures, podcasts, text-to-speech documents and read-aloud now use a voice set to your language in 30+ more languages.
Czech, Polish, Dutch, Greek, Hebrew, Thai, Persian, Urdu, Ukrainian and the Nordic languages are no longer guessed or misread.
Non-English podcast and video lecture scripts now translate technical terms and drop English filler like "okay" and "exactly".
Numbers and units are written as spoken in your language, and an English field term is said once after the native term.
Each video, podcast or document keeps one voice from start to finish.

"Ready" emails now preview what you made

The email sent when a long creation finishes now includes a preview of it.
Research emails lead with a TL;DR, confidence, and the files it created.
Video lecture and podcast emails show the poster or cover with timestamped chapters.
Slides and story books show their first pages, and flashcards and quizzes show a few cards.
Essays show their opening and outline, and files, transcripts, and narrations link straight to the download.

Smaller changes

Video lectures narrated from your own script now keep the script's language for the voice, title and chapter headings.
When your microphone can't be found, dictation and voice mode now show step-by-step help for your browser and operating system.
Photos picked on iPhone and Android upload more reliably, and you're asked to pick a photo again if the phone no longer shares it.
Finished video lectures are no longer lost to a brief storage outage at the final step.
Teams checkout now tells you exactly what to fix when it can't start.

October 2, 2026

Password, expiry, and revoke on shared links

A shared link can now require a password before it opens.
Links can expire after 24 hours, 7 days, or a time you choose.
You can revoke a link so the old address stops opening.
You can turn off saving and export for people using the link.
People you invited, and your team, still open it the way they already do.

Premium weekly AI Creation Credits are now 15

Premium now includes 15 AI Creation Credits each week.
Every Premium for Teams seat gets the same 15 credit weekly allowance.
Credits still reset every Monday at 00:00 UTC.
Credits are shared across flashcards, podcasts, video lectures, slides, and the other creators.
Laureate stays at 80 AI Creation Credits each week.

Choose monthly or yearly Premium

Personal Premium again offers Monthly and Yearly billing on the pricing page and in the plan picker.
Yearly is selected to start, and you can switch to monthly before checkout.
Teams seat prices are unchanged.

October 1, 2026

Buy credits for team members

Owners and Admins on paid Scholarly Teams workspaces can buy a credit pack for a member who runs out of weekly AI Creation Credits.
Buy credit packs from Team → Overview or Team → Billing, and they are charged to the team bill.
Purchased credits go to that member, never expire, and are used after their weekly allowance.
The credits card shows each member's extra balance.

GPT 6.1 Sol replaces GPT 6 Sol

GPT 6.1 Sol, OpenAI's newest mid-tier reasoning model, now powers the GPT Sol slot on paid plans.
It is used for Research, Slides, Video, Infographics, Podcasts, and customization profiles.
The AI Creation Credit cost stays the same.
Creations that had GPT 6 Sol selected switch to GPT 6.1 Sol automatically.

Arabic and Hebrew video lectures read right to left

AI Video Lecture titles and headings in Arabic, Hebrew, and other right-to-left languages no longer show reversed words.
Sentence punctuation now lands on the correct side.
Slides lay out right to left: headings start from the right edge, and steps and timelines run right to left.
Icons and markers now sit beside their text.

Smaller changes

AI Video Lecture title cards no longer add an "AI Video Lecture" label when the Scholarly logo is off.
Lectures in languages other than English no longer show English scene labels such as "Key idea" or "Definition".
Very large scanned PDFs, like full textbooks, no longer run out of memory when made into flashcards, slides, video overviews, or podcasts.
If PDF rendering fails, the message now asks you to try again instead of suggesting your file is unreadable.
Fixed a connection issue affecting Claude generation, chat startup, large document conversions, and URL imports.

September 30, 2026

Team credit use and billing at a glance

Team Overview shows AI Creation Credits used this week, when they reset, and a meter for every member.
Members get a Near limit or At limit tag before anyone is blocked.
Team Billing shows seats in use, open seats, and pending invites, and owners can add seats.
Failed payments and plans set to end are flagged with a link to Manage billing, and owners get one email when a renewal fails.
Paid Teams seats at the weekly limit can open Usage and buy a one-time credit pack, and Laureate subscribers get the same option.

Video lecture export packs

Owners of a video lecture can download one zip with the video, slide deck, narration script by chapter, and captions.
Captions come as both VTT and SRT, with a short README listing the sources.
The export pack is in the lecture's Export menu and beside the slide deck.
After buying Scholarly Teams, the workspace opens with a welcome card that starts a training video from your document.
One-person teams get a delivery checklist: create a lecture, download the export pack, set voice and look, then invite someone.

Long chats load as you scroll

Chat opens with recent messages and loads older history as you scroll up.
Long conversations use less data, and your reading position stays in place.
If loading history fails, you can retry without leaving the chat.

Smaller changes

PDFs, including uploads and generated decks, study guides, and reports, now start loading immediately instead of sometimes taking minutes.
Deleting a meeting that is still recording now says so, instead of showing a processing failure and a retry that could not work.
Home now keeps all running creations and failed tasks visible after you start a fourth creation, on desktop and phones.
Onboarding write-ups now join the team library even if checkout finishes first, and new teammates see what's already shared.
The voice-and-look step stays open until you edit the brand kit, and the sign-up page after Start a team explains the setup that follows.
People now flags invites whose email couldn't be delivered, and the Billing chip reads Payment failed alongside the banner.
Premium and Laureate Teams members no longer see Free-plan banners, popups, or credit meters, and see their real weekly allowance.
Limit buttons for team members now open Usage, and Usage shows weekly AI Creation Credits on Premium plans.
The plan picker shows the plan your team already includes, and team permission errors on custom designs are explained inline.
Removed members can no longer rejoin with an old join link.
The sole owner of a workspace without a team plan can now close their workspace to join another team.
Teams billing now holds up when Stripe events arrive late or out of order, and a canceled checkout no longer freezes invites.
The billing portal now returns you to Team billing.
Included weekly credits are now used before a purchased pack, and team usage shows real AI Creation Credit totals.
Canceling a video lecture or podcast before it starts now returns its credits.
The Video action on a video lecture now starts a new lecture from the original sources instead of failing.
The video create screen now lists PDFs that are still processing instead of hiding them.
Chat keeps original Word and PowerPoint filenames in its source list.
New video lectures show source page ranges from the pages actually read, and phone video controls share one labeled menu.
Quiz practice explains the account-wide answer allowance separately from the deck's question pool.
The account menu shows loading or Retry while your profile is arriving, and desktop Welcome dialogs fit shorter windows.
Slides and Chat better separate exact source identifiers from translated explanations and respect revision rules.
Slides and Chat now prioritize an explicitly requested slide count.
Stalled document previews now offer Retry.
Task pages explain delays once the usual time window passes, and update notices can be dismissed.
Switching accounts now cancels loading of the previous account's content and tasks.
Used-up Chat limits no longer show a competing model teaser, and video logo removal explains the relevant Premium feature.
Video question generation now uses the creator's checkpoint instructions, and short Podcasts use a compact chapter plan.

September 29, 2026

One AI Creation Credit per meeting, notes included

Each meeting costs one AI Creation Credit at any length, taken when you press Record and covering both transcript and notes.
Notes are written as soon as transcription finishes, and meetings recorded before today keep Generate notes at no cost.
Every plan keeps every meeting, and a spent allowance blocks the next recording instead of deleting an old one.
Free transcribes up to 10 minutes per meeting and paid plans up to 10 hours.
A meeting that fails to process, is abandoned, or is deleted before its notes exist gets its credit back automatically.

Auto (Best) now runs on Claude Sonnet 5.5

Auto (Best) for research, slides, video lectures, podcasts, and infographics now uses Claude Sonnet 5.5 instead of Claude Opus 5.5.
Auto (Best) still costs 3 AI Creation Credits and wrote scripts of the same quality in side-by-side tests.
Claude Sonnet 5.5 replaces Claude Sonnet 5 in the model menu and costs 3 credits when picked by name (Sonnet 5 cost 2).
Saved Sonnet 5 choices carry over automatically, and Claude Opus 5.5 can still be picked by name.

More complete creations, fewer premature stops

Creations no longer stop early based on spending, token totals, or tool counts, while processing deadlines and output checks remain.
Illustrated lectures check narration pace, Auto length plans for longer sources, and Math videos narrate the whole requested lesson.
Standard lectures can revisit source evidence and correct individual narration takes without redoing the rest.
Research keeps usable reports regardless of wording or citation counts and saves inline generated images.
Flashcard study questions keep answer keys and allow paraphrased evidence, and podcast companions keep longer text.

Smaller changes

Fixed a sorting error when loading libraries with many research reports; create screens now offer Try again on partial loads.
Create screens have outline-free settings cards and a flush action bar, and section headings use sentence case instead of ALL CAPS.
Slides and AI documents now receive your original request wording at generation start, helping prompt-only creations stay on topic.

September 28, 2026

Files as pills across chat, uploads and research

Files and linked sources appear as the same pill in chat, Add content, and Deep Research, showing their type and size.
Uploads show progress inside each pill instead of a blocking “Uploading” screen, and failed files show why with Retry.
Drop files anywhere in chat, on Home's Chat tab, or onto the Deep Research composer to attach them without a pop-up.
Download files from their pills, preview ZIP contents, and ask chat to send a ZIP bundle back.
Files over your plan's size limit stay listed with how far over they are, next to the upgrade option.

Failed creations explain why and offer the right fix

Slides, AI documents, study guides, mind maps and other creations now name the cause of a failure on Home and the progress page.
A failure caused by your sources offers Adjust, which reopens the creator with your original sources already selected.
The reopened creator explains why the last attempt didn't finish above Create, and suggests fewer pages when a PDF was too large.
Research and PDF summaries show a panel with the cause, whether your credits came back, and Try again when it can help.
Failed Image Studio images show why they failed and offer Try again.

Pages load without jumping around

Home, the Library, folders, chat, Settings, Teams, creators, flashcards and document pages show one spinner, then the finished page.
Counts no longer read zero first, and buttons and sections no longer flicker or drop in while a page loads.
Projects open to the whole workspace at once, with Recently generated, list controls and every row arriving together.
Project and page chats open straight to your recent chats or conversation, without a blank starter screen first.
Projects appear in the sidebar right after a reload, and toolbar and chat buttons stay put during background refreshes.

Smaller changes

The research page's Summary (was Results) now reads as a brief: the answer, how confident it is, and what is still open.
Tap an open question on the research page to ask it in the research chat.
Sources (was Evidence) leads with the sources the report cites, exports and citations share one menu, and citations copy to clipboard.
Research reports now open fully formatted instead of flashing plain text first.
The “How was this?” rating now appears in the page next to what you're rating, instead of floating in the corner of the screen.
Some new videos and podcasts use a different narration voice; rate how it sounds with thumbs up or down. Existing audio is unchanged.
Podcasts now build episodes from short or scanned sources, or from your typed topic, instead of failing with “no usable content”.
Free podcast gifts no longer use a document Scholarly generated as your source material.
Fixed dragging a video's progress bar breaking the player, and the loading bar now shows what's downloaded around your position.
Starting a tool from a file picked before signing up no longer stalls when your browser is slow to save it.
PDF study guides, cheat sheets and reports no longer fail when the typesetting service is briefly busy.
A failed quiz now reopens the quiz creator instead of flashcards, and quiz plan limits show the upgrade path in the creator.
Opening a creator with sources already chosen, such as “Make slides” from a document, now keeps those sources the first time.
Creations started from the upload window keep their failed row on Home, and podcasts in progress link to the standard progress page.
On phones, recovery actions are full-width buttons and the Image Studio remove button is reachable.
Deep Research's Start waits until every file is ready, and linked library items, Google Drive files and evidence types show as pills.
Chat can save images directly and keep working on larger requests without arbitrary page-read or image-attempt cutoffs.
Dragging files onto the Add content window now shows one clear drop area.

September 27, 2026

Extended research and Go deeper

Turn on Extended in the research composer to have a team of researchers, a scout, a fact checker, and a reviewer work on one question.
An analyst builds calculations and charts when the question needs them.
Extended research costs 3 AI creations, takes about 10 to 20 minutes, and asks you to confirm before it starts.
Extended research is available on paid plans; standard research still costs 1 AI creation.
Go deeper runs Extended research on any finished report, and you can ask for either in chat.

Delete and restore inside projects

Every item in a project now has Delete in its menu.
Deleted items stay openable from the project's new Deleted section and from Library → Deleted.
Bring deleted items back with Restore or the Undo toast.
Ask chat to delete or restore items in bulk, such as “restore what I deleted this week”.
Deleting is always reversible; nothing is permanently removed.

A clearer, better-sourced research workspace

While research runs, you can see the current step, time taken, source and fact counts, and the latest sourced facts.
The research composer fits on one row, with output, evidence, depth, model, and project in a single Options menu.
Evidence opens on the sources the report actually cites, ranked by how often they are cited.
Each fact keeps its exact source page and date, so citations point to the right source and newer figures win.
Research reads the actual page when an exact figure, table, or rule matters, and checks its report before finishing.

Smaller changes

Chat keeps short working notes for each conversation, like shortlists, figures, deadlines, and decisions, so it does not ask twice.
Details you ask chat to keep for one conversation stay in that chat instead of becoming a saved memory.
Extended research runs show what the fact checker confirmed or corrected, and research steps show who did each step.
The research page now shows your projects above Recent.
The report switcher and resume chip no longer overlap, and Extended and Go deeper reports are labeled.
Research now recovers pages that used to fail to load.
Create quiz has a new Write your own tab (also at New → Blank → Quiz) that is free and does not use AI credits.
Blank → Table opens a grid where you can type or paste from Excel or Google Sheets, free and without AI credits.
Worksheets, timelines, reports, SOPs, lesson plans, case studies, outlines, and flowcharts are now formats in the AI Document Creator.
The New menu is simpler, and the AI Study Pack and Paste text shortcuts are removed (you can still paste text from Upload).
Choosing Page from the New menu or a project now opens a blank page right away; use Draft with AI to have AI write it.
Chat answers that search the web now show a single Sources button instead of a second row of site icons.
Podcasts with the visual companion on now use their first chapter slide as the cover.
Fixed Claude Haiku 4.5 messages in chat being answered by a different model since mid-September.
PDF chat reads a run of consecutive pages in one step, so questions about a section or chapter answer faster.
If finishing Team setup cannot be saved right away, Scholarly now completes it the next time you open the app.
On the Free plan, the Premium tip appears above the chat box again after your first message.
Fixed delayed automatic credit refunds after failed AI creations and an error that could stop recording pages from opening.
Going back after connecting Google Drive no longer shows a false connection error.

September 26, 2026

Scholarly feels instant

Uploads and finished creations now appear in your projects right away.
Renames, deletes, moves, and settings changes apply instantly.
New chats open immediately, and pages you revisit open without a loading screen.
Home shows your last visit while it refreshes.
If a change can't be saved, it's undone and you are told.

Send a follow-up without waiting

Sending a message while chat is answering now stops that answer at once and starts on your new message.
The part of the answer already written stays in the conversation, and follow-ups are no longer queued.
Stop takes effect immediately, and you can type your next message right away.
Chat answers now stream more smoothly and start sooner.

Write a flashcard deck from scratch

Create flashcards has a new Blank deck tab where you write questions and answers side by side.
Paste a whole list with Import from text, then reorder, swap sides, and shuffle before you create the deck.
Blank decks are free and do not use AI credits.
Your draft is saved on this device as you type and stays until you create the deck or discard it.
The flashcard editor uses the same cleaner cards, with hover actions and an Add card row at the end of the deck.

Smaller changes

The scholarly.so home page now opens on the Scholarly composer: pick an output to preview it, and your choice carries through sign-up.
Choose Auto, editable Word, or PDF as your document format in Chat settings; your account remembers it across devices.
Start Recording on Home now shows a spinner during setup, ignores repeat presses, and opens the recording page sooner.
Fixed the “ready” notice for a finished creation not appearing while viewing a similarly numbered page.
New accounts now finish onboarding and see the welcome flow even if signup takes more than two minutes or they return later.
Word previews in narrow panels now keep the left edge visible and let you scroll across the full page.
File previews now adapt when you resize the window and switch to a full-screen preview on phones.
Older language-prefixed workspace links now open the correct page without changing your saved interface language.
Improved automatic recovery of generated PDF text and credit refunds after failed flashcard or research requests.
Retrying a page that has not saved yet now shows its progress spinner until the save completes.

September 25, 2026

Edit sent chat messages

Edit a sent message directly in its bubble and generate a fresh answer without leaving the conversation.
Earlier context and attachments stay, and version controls let you return to the original conversation.
Hover over a sent message to reveal Edit and Copy icons; on phones, these controls stay visible.
Editing older conversations keeps files and links that finished attaching after the message was sent.

Simpler model choices

The model menu now offers Auto, Auto (Premium & Laureate), and Auto (Best).
Auto (Best) uses Claude Opus 5.5 for research, slides, video, podcasts, and infographics.
Creation settings show a yellow warning with the exact multiple when a model spends more than one AI Creation Credit.
A new Use best model toggle in creation settings shows its credit cost before you select it.

Podcasts from your full source set

Creating an AI podcast now accepts every source your plan allows, not just three.
Paid plans can combine up to ten sources, and Laureate up to twenty, matching the other creation tools.

Smaller changes

Attach ZIP archives, datasets, audio, and other file formats in chat, which can inspect their contents (existing upload limits apply).
YouTube transcript extraction now says when the transcript service is briefly unavailable and invites a retry.
Password resets now sign out existing sessions, and emailed sign-in links have stronger single-use protection.
Private email invitations require a verified address, which you can verify from Settings → My Account.
Fixed infographic and mind-map PDFs being saved without the content needed for summaries and search.
The homepage, marketing pages, and upgrade screens now say Scholarly is used by 200,000+ people.
On a one-source plan, choosing a different file now replaces your pick instead of showing an error.
Links pasted without https:// are now accepted in every link field.
Pressing Enter to confirm Japanese input no longer sends a chat message, submits an answer, or creates a project.
Team invites explain invalid addresses and keep addresses beyond the 20-per-send limit ready for the next batch.
Team invites now link to seat management when you need more seats.
Empty library views offer a next step, and the source picker shows its loading state instead of an empty library.
The project picker now offers a retry when projects fail to load.
Quiz shortcuts no longer fire while you type a follow-up, and flashcard autoplay no longer swallows spaces in text fields.
Focus-mode Flag and Hint now report whether they worked.

September 24, 2026

Emails when your plan limit resets

If you reach a Premium or Laureate plan limit, Scholarly now emails you when that allowance resets.
Manage these emails separately in Settings → Notifications → Plan limit resets.

GPT 6 Luna for flashcard Auto on paid plans

Flashcard Auto on paid plans now uses GPT 6 Luna to create cards, refine them, and generate study questions.
You can still select Gemini manually for flashcards.
Video Auto on paid plans now uses high reasoning, matching the default for other paid creation tools.

Smaller changes

Improved generation of accurate diagrams, readable multilingual headings, and flashcard answers that keep important qualifications.
Fixed Prompt AI flashcards asking for uploaded lesson pages even when you supplied a valid study topic.
Video and flashcard pages restore saved tabs and cached content after loading, and explicit links still open the requested tab.
Lists, math, and reference links keep their formatting while chat answers stream, and code previews preserve special characters.
The chat message box stays in place below the loading indicator while earlier messages load.
Projects and Folders in the sidebar show their headings immediately instead of appearing after the page has painted.
Podcast and video lecture pickers no longer offer uploaded videos whose transcription finished without a usable transcript.
The podcast library lists only videos that already have a transcript, so Create does not stop on one still being transcribed.
Rapid Next presses no longer skip quiz questions, video lectures pause for timed questions, and long question overlays can scroll.
Files stay attached when you open More options, Office filenames stay recognizable, and video customization shows selected sources.
Dialogs keep keyboard focus inside them, and cookie preferences show your saved choice.
Progress no longer shows invented zero-second timings, and media duration and upload limits are clearer.

September 23, 2026

GPT 6 models and Claude Opus 5.5

GPT 6 Astra, GPT 6 Sol, and GPT 6 Luna replace the GPT 5.6 family.
GPT 6 Luna is the new default model for chat and every creation tool, and GPT 6 Sol is the standard paid pick.
GPT 6 Astra is the new Frontier model at 3 AI Creation Credits per creation.
GPT 6 Astra works for research, slides, video lectures, infographics, customization profiles, and now podcasts.
Claude Opus 5.5 replaces Claude Opus 5 as the other Frontier model, and saved model choices carry over automatically.

A calmer Page editor with faster autocomplete

Pages open as a quiet writing surface with a large title, a comfortable reading column, and larger body text.
Selecting text offers bold, italic, underline, code, and Send to Chat in place, and Export sits at the end of the formatting bar.
Autocomplete suggestions arrive about twice as fast from a new model and stay on screen while you type the predicted words.
Tab accepts a whole suggestion, Cmd or Ctrl and the right arrow accept one word, and a tap or click works too.
Chat edits appear on the page with removed text struck through and new text highlighted, decided by one Accept or Reject bar.

Meeting transcripts first, notes when you need them

Recording automatically saves a transcript, and Generate notes adds a summary, decisions, action items, and timestamped topics.
Paid plans include up to 5 hours of transcription per meeting, and notes cost 1 AI Creation Credit when they are ready.
Free includes one saved meeting with up to 30 minutes of transcription and AI notes, and deleting it makes room for another.
Failed notes can be retried without buying transcription again, and pricing pages now list meeting limits for each plan.

Smaller changes

Long chat answers keep finished paragraphs and code blocks in place while the rest streams, so streaming is smoother.
Pages opened from a link, bookmark, or refresh now show their title and content right away instead of a loading spinner.
Renaming a page, changing its icon, or accepting a chat edit now updates the sidebar, library, and search right away.
Signing in with an unknown email offers to create an account, and signing up with an existing one offers to log in, keeping the email.
Pasting a YouTube or web link into the text box now adds it as a link instead of showing an error.
Private, deleted, or unavailable YouTube videos and empty playlists are flagged when added, and playlists show which video will be used.
Chat dictation and voice mode use your chosen recording microphone, report when no sound arrived, and show browser steps if blocked.
Voice mode now creates a chat only after the microphone opens, so canceling or blocking it no longer leaves an empty chat.
On phones, the model picker fits beside the microphone button without covering it, even with a long model name.
Connecting Google Drive now says to tick the Google Drive box and what was missed, and brief hiccups no longer read as an expired code.
On Free, pressing Record with your meeting slot full lets you open or delete that meeting, or upgrade, before anything starts.
Canceled tasks show whether their credit was returned or used, and tasks that never used a credit no longer say one was restored.
A failed research report no longer blocks starting a new one while its credit is being settled.
Run now on a scheduled flashcard deck, video lecture, or research report starts right away instead of waiting for the original time.
Retrying older generations, meeting notes, or PDF summaries is no longer canceled as "stalled" moments after it starts.
Taking a minute or two to answer a research session's clarifying questions no longer fails the session.
Work interrupted by a server restart resumes sooner, and a failed PDF text extraction can be retried instead of blocking its summary.
Canceling a video lecture now stops its generation immediately.
Team and shared project descriptions and instructions now update instantly, and rejected edits restore the saved version with a reason.
Video lectures, infographics, and document generators now accept up to 10 sources on Premium and 20 on Laureate, instead of 3.
Dictation clips with no usable audio now say so right away and ask you to record again instead of offering a Retry that would fail.
Tabs running an older version of Scholarly after an update now tell you to refresh instead of showing raw web-page code.
New pages show one hint and a Draft with AI link, and the page title no longer overlaps the first line.
Projects have a calmer layout with quieter tabs, the most-used create options in one row, and descriptions you edit in place.
Project renames update everywhere on Enter, and the save indicator only appears when you are offline or a save fails.
Chat panels open with one short line and suggestions that fit the page or project you are in.
Video lectures have a "Use my script as written" option that keeps your script's wording and order (free videos stay 2 minutes).
The recording bar has a microphone button to pick an input before starting or switch mid-meeting without stopping the recording.
The recording page now says right away when the microphone is blocked, with steps for your browser, instead of erroring on each Record.
If a phone or in-app browser refuses the microphone, the recording page says the live transcript is degraded instead of "listening".
When a recording picks up no speech, you can switch microphones and continue into the same recording in one step.
Recordings with unreadable audio are rejected once with advice to re-record, instead of retrying and failing with a generic error.
New personal projects open right away, so naming, adding pages or folders, and editing instructions no longer wait for the save.
Pending sync changes are backed up in this browser and resume after a reload, and project workspaces show save and retry status.
While a generation runs, its progress page suggests what to make next from your materials, never repeating the one in progress.
The create modal's waiting screen has a View progress button next to Close to follow a generation step by step.
Output languages named in your prompt now override the language menu for videos, podcasts, documents, slides, infographics, and mind maps.
Slides, study guides, cheat sheets, and other outputs now cover every major section of long sources and every topic you list.
Exam grading accepts equivalent answers such as 0.5 for 1/2 and alternative wording, and grading outages no longer count as wrong.
Mind maps keep every label in full, get a larger canvas when needed, open fitted to the screen, and can be dragged to pan when zoomed.
Generated videos are encoded to stream smoothly and seek quickly, so playback no longer stops every few seconds on slower connections.
Video lectures made from a PDF now show the document's real figures, charts, tables, and diagrams instead of text cards.
On iPhone and Safari, slow-starting videos show a loading spinner, and tapping starts playback instead of pausing.
The flashcard Questions tab counts progress across the whole deck, questions can be read aloud, and they stay within what your cards say.
Free accounts see the automatic upgrade popup at most three times, and the home banner no longer covers the plan window.
Downloads remember your last format, CSV files open correctly in Excel on Windows, and non-English titles stay in filenames.
Opening and leaving a project quickly no longer crashes the project page.

September 22, 2026

Recordings warn you when no speech is heard

Finishing a recording with no speech now asks what to do instead of failing minutes later.
You can check your microphone and keep recording into the same recording, discard it, or process it anyway.
The capture bar and recording page now flag a live microphone with nothing transcribed for 20 seconds.
A recording that fails because no speech was found now says so on its page, with playback-and-retry advice.

Grok 4.7 and GPT Image 2.5

Grok 4.7 from xAI replaces Grok 4.6 on research, slides, video lectures, podcasts, infographics, and customization profiles.
A saved Grok 4.6 choice continues on Grok 4.7, which is not available in AI Chat.
Image Studio adds GPT Image 2.5 Flare and GPT Image 2.5 Sunburst.
Flare replaces GPT Image 2 in Image Studio and for images generated in chat, flashcards, slides, and other tools.
Sunburst is a slower precision model for edits that should change only what you asked.

Long iPhone videos upload reliably

QuickTime (.MOV) uploads already in H.264 are repackaged as MP4 in seconds instead of re-encoded.
Longer recordings get a conversion window that grows with the file.
If a conversion still fails, the message says whether the clip is too long or the media service was briefly unavailable.

September 21, 2026

Video lectures recover and finish more reliably

A lecture that stalls before anything is rendered now hands the work to a second model instead of failing with nothing to show.
When a scene added mid-lecture has the wrong shape, the error names the exact fields to fix so the lecture corrects itself in one step.
Lectures keep going with the current scenes when the plan review adds or replaces scenes while the first chapter is recording.
Narration that reaches the 45-minute ceiling now renders every voiced scene instead of discarding the whole recording.
Failure messages for the remaining cases now say what actually happened.

Automatic recovery for interrupted creations

Temporary generation failures are retried automatically up to three times.
Recovery reuses saved output and completed steps.
Tasks show recovery progress and credit settlement status.
A connection problem no longer incorrectly marks a creation as failed.
Recording uploads are confirmed byte-for-byte, so a retried or duplicate upload can never lose or double audio.

Refreshed help center

The Notes and Pages guide is now one Pages guide covering a blank page and an AI-written first draft, and old links still work.
New guide: Continue Where You Left Off, on how your place is kept in lectures, podcasts, PDFs, pages and decks across devices.
New guide: Following a Creation While It Runs, covering the live progress page, what each step means, and how to cancel.
Guides for video lectures, podcasts, Home, uploads, PDFs, billing, Explore and more were updated for this month's releases.

Smaller changes

The writing feature page no longer advertises a retired shortcut and AI Edit popup, and the video and podcast pages were updated.
Text blocks have their options menu back on phones, and you can now delete a block from that menu on phones and desktop.
The 10-second skip buttons are back on the podcast episode page on phones.
In chapter lists, the row you are hovering is no longer shaded exactly like the chapter that is playing.
Fixed the project workspace showing a blank error page when your session refreshes; it now recovers on its own.

September 20, 2026

Video lectures run the length you pick

A video lecture is now built to the runtime you choose, in every mode.
Narration pace is measured from the first chapter, and the remaining chapters are sized to match.
A 15-minute pick no longer arrives as a 6-minute video.
Short plans are extended with new source-grounded scenes instead of being cut.
No video is refused for being long or short.

Faster, more visual Standard video lectures

Standard video lectures start recording the opening chapter as soon as the scene plan is ready.
A 12-minute lecture reaches its first narration in roughly half the time it used to.
The progress page shows narration and rendering as separate steps, with counts of scenes planned and recorded.
Long lectures now use many short scenes, around three a minute for a 45-minute lecture.
Scenes that run long reveal content in stages timed to the narration, so the screen changes every 10 to 15 seconds.

Public library and blog archive

A new public library at /explore/library lists every publicly shared item by title across numbered pages.
The library includes pages, flashcard decks, videos, podcasts, PDFs, research reports, and chats, each opened directly.
The blog has a paged archive (/blog/page/1 through /blog/page/18) covering all 839 posts, newest first.
"Keep reading" under a blog post now shows the posts published just before and after it in the same topic.

Smaller changes

Read-aloud now plays at its original volume, keeps speech intact across audio chunks, and plays flashcards more smoothly.
Temporary voice-service errors in video narration are retried more reliably, and completed audio is reused.
Video lecture text is larger and sized to each scene, light-mode cards have more contrast, and repeated cards appear once per chapter.
If a video lecture's workspace is replaced during packaging, the lecture is now retried and your credit returned.
A momentary network fault while indexing a PDF is now retried instead of discarding the whole document's indexing.
Exceeding the three-source limit (mind maps, slides, and other formats) now returns you to your source list with the reason shown.
Japanese pages link directly to the English-only blog, help center, and topic hubs, whose language switcher opens the Japanese home.

September 19, 2026

A refreshed Page editor

The AI Essay Writer is now Page, with an always-visible, centered formatting bar and a compact mobile toolbar.
Choose Page on Home, then Start blank to open a new page right away, with no prompt or setup step.
Autocomplete suggests a continuation after you pause; accept it with Tab or the Accept button, or dismiss it instantly.
AI Chat changes now appear drawn on the page with one Keep or Discard, and the floating AI Edit popup is retired.
Long pages stay responsive while you type, and chat uses the latest page content when you send a message.

Redesigned podcast player and chapter list

The player bar is one fixed-height row, with captions, errors, and the resume note floating above it.
The progress bar has one segment per chapter; drag anywhere to scrub, or click a chapter's edge to start it.
Chapter list rows stay quiet until hovered, and the playing chapter gets its own indicator and progress line.
On phones, the player bar keeps the episode title readable instead of truncating it to a few characters.

Cancel flow shows your options before ending renewal

Canceling from Settings → Billing again shows what you would lose and asks why you are leaving before renewal is turned off.
You see a matching offer: a pause, a discount, an annual switch, or a free artifact built from your material.
The one-click Cancel renewal button from September 8 has been removed.

Smaller changes

Code paste and image captions in Page save reliably, and undo restores deleted blocks in their original positions.
Page tables reflect incoming edits, and shared read-only pages keep their tables, code, and captions locked.
Page icons sit beside the title, blank lines keep their spacing, and the Export menu opens without a loading spinner.
The Page editor now opens ahead of its chat sidebar.
Page spelling and grammar checks now run only when requested, and autocomplete respects your settings.
Video lectures whose workspace stops responding after rendering are now rebuilt in a fresh workspace instead of failing.
Claude Haiku 4.5 chats with reasoning turned on now reply on Claude instead of erroring and switching to another model.
Long Claude Opus 5 steps in AI generation are no longer cut off after five minutes and retried from scratch.
Canceling a slide deck, document, or infographic no longer shows an error when the cancellation already went through.
A canceled Best-quality generation now returns its daily Best slot.
Double tapping Record, Chat, or an upload option in the phone welcome flow no longer starts it twice.
Continue where you left off on Home now shows uploaded PowerPoint and Word files by their original names, not the .pdf name.
Videos stuck buffering for 30 seconds now offer a Retry action that reloads from your place.
PDFs with very long file names are no longer dropped as sources in AI Video Lecture, slides, and document generation.

September 18, 2026

Cancel creations from the progress page

You can cancel supported creations from their progress page after confirming.
Narration and transcripts can be canceled while they are queued.
The progress page tells you when a creation is taking longer than usual, with a cancel action where supported.
Canceled work is clearly labeled and no longer shows failure-recovery actions.

Fewer lost study guides, lectures, and slides

Study guides and documents are kept if search indexing hits a temporary provider error, and indexing retries in the background.
Slideshow-style video lectures no longer fail while trimming a video tail that ran past the narration.
Lectures that switch models mid-run go to the right render step instead of stalling.
Slides and documents can recover from repeated empty AI responses using another supported model, keeping completed work.

Smaller changes

Flashcards imported from Quizlet, Anki, or a file now open with study mode ready instead of briefly showing a page-not-found message.
On Safari, files you selected before upgrading are kept through checkout more reliably.
Jumping to a flashcard by number opens its question without revealing the answer.
Saved playback and reading positions now belong to the specific generated file, so replacement content starts at the beginning.
Word and PowerPoint sources keep their original names in the viewer, sharing, and title editor.
Renaming a source keeps its rendered PDF copy consistent.
Older PDF bookmarks with known zero-based page numbers are converted once; ambiguous ones start at the beginning.
Closing a PDF while its saved page is loading no longer overwrites that page.
Public videos pause when the save prompt opens and stay at your place until you press play.
Signed-out viewers of public videos keep progress on their device without failed account-sync requests.
When your upload allowance is exhausted, Upgrade is available before you select a source, and Library or Drive sources still work.

September 16, 2026

A clearer first run on your phone

The phone welcome steps now use the same tap rows, headers, and file-size note as the rest of the app.
If you have nothing to upload yet, you can record on the spot or start chatting and add files later.

Pick up where you started after signup

Files and links selected in our public tools now carry through signup, so you don't have to select them again.
The welcome action screen keeps all its choices after signup.
Recording is shown first when you arrive from a lecture recorder.
Your material is shown first when you start a source-based creation.

Smaller changes

Math video lectures whose narration runs past your chosen length now render the scenes that fit instead of failing.

September 15, 2026

One Recent list on Home, grouped by day

Uploads, AI creations, chats and folders now share one Recent list grouped into Today, Yesterday and Earlier.
In progress and Scheduled sections appear at the top when work is waiting, with Run now on scheduled rows.
Continue where you left off is now a compact strip of cards showing minutes remaining, the page you stopped on, or cards reviewed.

A live progress page for every AI creation

Every generation opens a progress page with steps (reading sources, planning, rendering, packaging), elapsed and typical time.
Generations keep running if you close the tab, and you can return to them from Home any time.
Agent-built creations (video lectures, slides, documents, spreadsheets, research) show turns, tool calls, subagents and the model.
Running and failed tasks open their progress page, with retries, upgrades and the reason something stopped.

Podcasts keep playing while you work

A mini player with play, skip, speed and the scrubber stays at the bottom as you open sources, chat or browse your library.
Podcasts and video lectures now work with your lock screen, notification shade and headset buttons.
Playback speed, volume, captions, PDF zoom and your last used tab now follow your account across devices.
Video lectures gain picture-in-picture and the J and L keys to jump ten seconds back or forward.

Smaller changes

New flashcards and quizzes use the standard creation flow again, without the experimental 10-card preview.
The Free chat limit now offers a direct upgrade to keep chatting and explains that the free allowance does not reset.
Some Free accounts can now upload files up to 32 MB, and the upload picker shows your current limit.
Google signup, including One Tap, no longer fails when Google does not share a profile name.
Pages, research reports and flashcard decks reopen where you stopped on any device, with a Start over option.
Create modals reopen with your last used settings, with a Reset to defaults. Custom instructions are still never saved.
Deleting or removing something from your library now shows an Undo that also restores it to its folder.
Search (⌘K) remembers your recent searches, and you can clear them.
Fixed AI Meeting Notes on iPhone, iPad and Mac Safari dropping up to the last 10 seconds when you press Finish.
Fixed full ten-minute voice dictations in Safari being rejected as too large.
Video lectures, podcasts, infographics and documents now warn right away when a selected PDF is blank.
Fixed RTF files that Windows labels as Word documents being reported as corrupted on upload.
Stalled uploads and flashcard requests now fail after a minute so you can retry, instead of after five minutes.
Fixed progress bars missing on items past the first 100 of one kind in long library lists.

September 14, 2026

Continue where you left off

Video lectures, podcasts and PDFs now reopen where you stopped, on any device you sign in from.
A small note shows where playback or reading picked up and offers Start over.
Anything you finished still opens from the beginning.
Home shows a Continue where you left off row with up to four unfinished items, including where you stopped and how long ago.
Library, folder and project rows show a thin progress bar for partly watched, listened to or read items.

DeepSeek V4.1 Flash for chat and AI creations

DeepSeek V4.1 Flash is available under More Models at the standard credit rate.
Use it in Chat, Deep Research, Slides and documents, Video Lectures, Infographics and Mind Maps, Flashcards, and Podcasts.
It can also be selected in Customization Profiles.
The model understands images as well as text and supports tools and optional thinking in chat.

Video generation keeps its progress

Video generation can now recover interrupted rendering connections.
When an eligible fallback model is available, a stuck run continues using the narration and files already created.
Recovery keeps the original time limit and avoids repeating completed work.

Smaller changes

Fixed blank pages and spreadsheets created in a project not saving to it or appearing in its Content list, including after connection failures.
Chat now retries briefly when its file workspace is busy, so file and code tools stay available for the rest of your request.

September 13, 2026

Checked citations and Sources button in chat

Every citation in a chat answer is checked against your linked sources before the answer is saved.
Citations to a source the chat cannot see, or to a page the document does not have, are removed.
Answers that cite something have a Sources button listing each source used and the pages it cited.
Opening a source from the Sources list takes you straight to the cited page.
Thumbs up and thumbs down are back on every assistant answer, and tapping your rating again clears it.

AI Chat model changes and faster tool use

Claude Sonnet 5, Grok 4.6, GPT 5.6 Terra, Gemini 3.1 Pro Preview, Kimi K3, and Mistral Medium 3.5 are no longer offered in AI Chat.
Existing chats using these models switch to your plan's default chat model on the next message.
Model availability for other AI creations is unchanged.
Chat loads extra tools only when needed and reads large files in smaller sections, so multi-step requests deliver results sooner.

Smaller changes

If chat's code workspace is at capacity or fails to start, the assistant can explain and retry instead of showing an unclear error.
Tab and Shift+Tab navigate normally in the flashcard editor; answers come from the Generate answer button, with a credit prompt on phones.
Choosing Page in the Blank section now opens an empty editable page directly.
Shared videos without a generated summary now show a clear unavailable message instead of a loading failure.
Next now accepts pasted text directly, including on phones, with the same source limits and validation.
Importing a Word or PowerPoint file opens the upload flow, showing your plan's upload limit and upgrade options instead of an import error.

September 12, 2026

Clearer video lecture details and Freestyle by default

The video page shows the requested length next to the delivered runtime, the source page range when available, and narration language.
Free previews are labeled with their runtime and the target length of the full lecture.
Paid plans now start new video lectures on Freestyle, where the AI picks the format and tools that suit the material.
Every style is still listed, and a style you pick yourself always wins.
Standard lectures get more room for script planning and an automatic retry, so dense sources follow your requested length more closely.

Listen to podcasts before they finish

For episodes with multiple chapters, you can start listening to chapter 1 while the rest is still recording.
When the full episode is ready, playback switches over and keeps your place.
Chapters are recorded two at a time, and preparation for the final audio mix starts while recording is still under way.

Challenge exam answer keys and get AI feedback

Each exam question can show a cited source passage when available, or the original flashcard answer, for review.
You can challenge any answer key once per attempt, and AI will uphold it, correct it, or report insufficient evidence.
Your score increases when a correction makes your answer right.
After an exam, ask for AI feedback on the whole attempt: what you got right, where you went wrong, and what to study next.

Smaller changes

Restored normal processing to cut wait times for videos, podcasts, research, slides, and documents, keeping your selected model.
Credit multipliers in AI creation model menus now stay aligned to the right of each model name instead of wrapping to a second line.
Scholarly now retries YouTube transcripts after a dropped connection and reports a temporary outage instead of missing captions.
Math video lectures made with GPT models use the vetted Manim animation recipes again instead of the slower free-form renderer.
If a video lecture's rendering workspace stops mid-generation, Scholarly replaces it and continues on the same model.
Fixed a narration handoff issue that could make video lectures repeat a failed step and eventually stop.
Fixed a crash when opening podcast creation, including when restoring a failed episode's sources.
Interactive questions now refresh after you regenerate a video lecture at a longer length, so your answers save.
Standard lecture scenes change more often, equations fit within the frame, and labels wrap to keep longer content readable.
Paid plans apply the model or plan's reasoning level more consistently to video, podcast, study, PDF, flashcard, and exam tasks.
Podcasts created from files now get titles based on the conversation instead of the file name.
Failed flashcard runs now show a reason and recovery actions: try again, try another model, pick fewer pages, or replace the source.
Failed video lectures and podcasts now explain why they failed and offer only the fixes that apply.
Free accounts now see upgrade options below PDF summaries and exam feedback; existing results stay until you regenerate.

September 11, 2026

Simpler cookie choices

The cookie banner now offers two choices: Accept or Deny optional advertising cookies.
Product analytics stays active with either choice.
You can change your advertising choice from Cookie preferences in the footer or Settings.

Clearer YouTube transcript errors

When a YouTube transcript can't be fetched, the transcript tool now says exactly why instead of a generic failure message.
Reasons include a deleted, private, or region-blocked video, a video still live streaming, or transcription still running.
The tool also tells you when the provider is temporarily unavailable.
Retry is only hidden when retrying genuinely cannot help.

Smaller changes

Signing out or switching accounts during a creation no longer carries in-progress items or notifications into the new session.
Fixed audio that was still processing reporting a failed upload after signing out or switching accounts.
Finishing a creation open in two places no longer adds a pop-up notification on top of the result already on screen.
Login, sign-up, and account recovery pages use a more compact desktop layout with consistent form widths and text sizes.
Improved page syncing when reusing saved content, and made project access checks consistent across valid project links.
Opening a finished quiz from its completion notice or task list now goes to the quiz page instead of the flashcard deck.
If your plan no longer includes expanding a flashcard preview, expand and retry now open upgrade options instead of an error.

September 10, 2026

A team library and reorganized team workspace

Opening your team shows a library of all shared sources and creations, newest first, including your own shares.
Filter by type with live counts, search by title or sharer, and see who shared each item, when, and its view count.
Work shared with your team also appears in All Content with a Team badge naming who shared it.
Team pages are now Library, AI profiles, People, Overview and Billing, and old Usage, Settings and Library links still work.
Team admins can approve or deny join requests from the People page, and approving sends an invite.

Team setup creates a shared project and first draft

Team setup asks what your team does, then asks for one source: a file, a link, or an item from your library.
Scholarly creates a team-owned project, shares the source, and drafts a first artifact such as an SOP, report, or case study.
Invite emails and the three-day reminder list the most recently shared items as links.
Teammates who join land in the shared library, and the welcome tour opens on what the team has shared.
Setup is rebuilt for phones with full-width actions and proper touch targets, and you can now move back between steps.

Team-owned projects

A project can now be owned by a team, so every active member opens it without an invite.
Team projects keep every member's contributions and stay when the person who created them leaves.
Transfer a personal project to a team, or hand a team project to one member, with a dialog showing who gains and loses access.
The project header names the owning team, and the gallery groups team projects above personal ones.
Project pickers name the team on shared destinations, so you can see when you're filing into shared work.

Smaller changes

Annual Premium for eligible new personal subscribers in Europe now equals six monthly payments. Existing subscriptions keep their price.
Accept or reject optional cookies or set analytics and advertising separately; Global Privacy Control keeps advertising off.
The cookie banner now appears only on public landing pages, not in the dashboard.
Updated privacy, service, and commercial-use policies covering content rights, AI processing, Google Drive access, Teams, and deletion.
Team usage now lives in Overview as Activity, and model access moved to Billing as Plan and AI models.
Join request emails open straight to the request, and running out of seats is explained inline with a link to Billing.
Library filters stay usable while more items load, and What's New no longer appears in the purchase welcome.
Team setup retries filing a source into the same project and shows an inline recovery message.
Transferring a team project to a person now also moves legacy folder content and removes the previous team's access.
Teams without a team plan that already used their free creation keep the new source and project, and see what upgrading would build.
The invite step shows what teammates will see and offers to share something first when the library is empty.
The share modal now shows who opened an item and when (once per person per day, excluding your own opens).
The weekly share activity email now includes teammate opens alongside public-link activity.
Testing a flashcard preview (up to 10 cards) with 20% of new Free accounts; existing accounts and decks are unchanged.
Upgrading expands a preview deck from its original sources, keeping edits, favorites, and progress. Retrying a failed start resumes it.
Removed the Unlimited practice promotional banner from the flashcard study view.
Loading spinners now match button text color, so they stay visible on customized buttons in light and dark mode.
When a creation can't read a source, the error now says how to fix it and no longer shows internal IDs.
Team projects shared with you by invitation stay in your project list, and editing team content requires an owner or admin role.
What's New is a compact corner popup again and no longer dims the page or blocks your work.
Personal Home, onboarding, upload suggestions, project-chat prompts, and sidebar navigation are back to their previous behavior.
Artifact headers again use the existing Share controls, without the added companion actions or Teams promotions.
Connection errors while preparing recorded audio no longer say the recording was deleted.
Signing out or switching accounts no longer reports pending creations as failed.
If voice input can't be transcribed, your recording is kept and a Retry button sends it again.
Starting AI Meeting Notes from the New menu now shows an error if it can't be created.
Error messages no longer show raw web-page code when a request fails.
When a linked YouTube video can't be transcribed, chat now says so and suggests asking again.

September 9, 2026

A redesigned New menu

The New menu is a one-screen grid grouped by outcome: Listen & watch, Present, Write, and Research & data.
Study formats and blank documents sit on one strip beneath the grid, with no scrolling needed.
Report, SOP, case study, spreadsheet, and timeline are now their own entries instead of options typed into the document creator.
Meeting notes now sits with the other work formats.
Upload and Import from apps sit together at the top, so adding a source is always the first option.

Scholarly rebuilt for phones

Home keeps the desktop order but runs edge to edge, with Upload, AI video lectures, AI flashcards and AI podcast as a 2×2 grid.
Every menu on a phone, including New, More tools and the model picker, now opens as a full-width bottom sheet.
Video, PDF, file and recording pages run the player or document edge to edge, with sections in a scrolling row of labeled chips.
Your library, folders and projects are flat rows, with show, group and sort in one sheet.
Settings is a hub of grouped lists, and tapping a group opens it with a back button like other phone settings.

Simpler Add content and creation on phones

Add content is a native-style sheet with rows for Take Photo, Photo Library, Choose Files, and a single Paste a link field.
From Library, Paste Text and Google Drive each open their own page with a Back button.
The project you're saving to is a row at the bottom, and the Next button only appears once something is added.
Customization is one screen: style, Language, voice, resolution, length, instructions, and one Create button.
Create asks before spending a free AI Creation Credit, so a stray tap can't use it up.

Smaller changes

Automatic upgrade popups no longer interrupt you on phones; plans still open from Upgrade or when you reach a plan limit.
The plans screen on phones shows the price and upgrade button without scrolling, with each feature on a single line.
The chat tool row on Home now fits the screen width, with tabs that don't fit moved to the top of More tools.
On phones, the podcast player dock is one row, and flashcard actions sit in one row under the card.
On phones, chat names the thread in the header, Research is a single readable report, and the page editor runs full width.
On desktop, the Add content sheet drops its subtitle and link badges but keeps the Upload / Library / Paste text / Google Drive control.
Reaching your free AI Chat allowance now shows what ran out and how to upgrade, instead of a generic error with a failing Retry.
Fixed Gemini chat turns that gathered information but wrote nothing; they now get a real second attempt at the answer.
Writing suggestions that delete a word now show a Remove button that works, instead of crashing the page editor.
Undo and Redo shortcuts work while typing in the page editor, including after applying a writing suggestion.
Explore search no longer matches embedded image data as a source or suggests related searches when nothing is found.
Sign-up has one accessible submit button, reliable Enter-key submission, and larger, more readable desktop fields.
The checkout recovery testimonial and Team navigation label now follow your app language, including Japanese.
Link previews now work for DOI, library link-resolver, and YouTube links, and no longer fail on long pages or sites that block readers.
Chat now says a turn was stopped safely when another attempt already owns it, instead of showing a generic error.

September 8, 2026

Editable Word, PowerPoint, and Google exports

Pages, generated documents, research reports and sources, videos, and podcasts can export to supported Word or PowerPoint formats.
Editable exports require a paid plan; page PDF options and research citation styles remain available.
Text exports are editable, while image-based slides keep their original visuals.
With Google Drive connected, save as editable Google Docs or Slides; otherwise the menu offers Connect Google Drive.
Exporting creates a separate copy and keeps the original sharing permissions.

Sort, group, and arrange your content

Sort Projects and All Content by name, when something was added or last worked on, or type.
Name sorting reads numbers in titles, so Chapter 2 comes before Chapter 10 and a podcast lands next to its chapter.
Group a list by type, by date, or into the sources you brought and the artifacts Scholarly made from them.
Choose Custom order to drag rows in a project; the order is saved, so people you share it with see the same arrangement.
Projects now have a search box and a type filter based on their contents, and both work in Shared with me.

Faster project workspaces

Large projects load noticeably faster because opening a project no longer scans your whole library.
Uploads of large recordings and files have more time to finish on slow connections.
All Content shows one initial loading indicator and keeps available content visible while projects load.
Stalled or failed project requests now offer a retry.

Smaller changes

The All Content toolbar is now a single row with search, location, grouping, and sorting, with content types on one line beneath.
Longer-media upgrade offers explain the target duration and continue the selected video or podcast after checkout.
The video page no longer shows the separate Create another version button.
File-size and PDF-page upgrade recommendations pick the lowest plan that fits your source, including Laureate for larger files.
Premium stays selectable, with a warning when its source limit is too low for your source.
AI Creation Credits are named consistently across plans and tools.
Exhausted chat and flashcard controls now show their disabled state.
Personal Laureate now offers monthly billing.
Selecting the allowed number of sources no longer shows a warning.
Adding more sources than allowed shows a persistent inline error with the limit and upgrade option, keeping accepted sources ready.
Typing now clears note placeholders immediately, and blank pages no longer show an unrelated upload warning.
Flashcard insertion labels now distinguish Add at top from inserting below.
Dark mode text is easier to read.
Quiz counts reflect prepared questions, and plan and recording explanations are clearer.
The homepage explains the workflow from your source material to answers, flashcards, video lectures, and podcasts.
Personal pricing explains which plan fits a first source, regular weekly work, or a heavier workload.
Sign-up clarifies that creating an account does not charge you and keeps your next step when you arrive with a plan selected.
The creation picker explains when to choose chat, flashcards, video, or podcast, with your sources and Back path still available.
Locked video and flashcard options, and custom exams past the free attempt limit, now open plan selection directly.
Onboarding and study follow-up emails now reach all eligible recipients.
New complimentary artifacts sent by email are podcasts, with a clear subject saying they are ready.
Fixed upgrade-offer emails not reaching eligible subscribers who had reached their weekly creation limit.

September 7, 2026

Updated help center guides

The help center now covers recent features, common troubleshooting steps, and team rollout.
If a guide is not helpful, you can optionally say whether it is missing information, out of date, or hard to follow.

Smaller changes

AI Chat now says when the image model declines a prompt on safety grounds and continues with the rest of its reply.
Blog articles that fail to load due to a brief connection problem now invite a retry instead of showing “Blog post not found”.

September 6, 2026

Scholarly remembers where you left off

Creation setup keeps each format's draft when you go Back, edit sources, or try another format.
Video, flashcard, podcast, study-guide, and infographic languages now apply to that creation without changing your account default.
PDFs, podcasts, and uploaded or generated videos remember your reading or playback position in this browser.
Audio and video players also remember your playback speed.

Clearer downloads, billing, and sharing

Paid downloads start directly, and failed PDF, video, or podcast exports can be retried.
Credit packs now show the price per credit.
Renewal-off status is clearer, and cancellation includes optional feedback, with alternatives available separately.
Sharing buttons name the audience change before you copy a private link.
Download dialogs open sharing options, and private QR links require choosing an audience first.

Smaller changes

Word, PowerPoint, or Excel files too complex to convert now fail right away and suggest a smaller file or a PDF instead.
Opening the Google Drive picker without a connected Drive account now explains this and points to Connections.
Mobile artifact chat now opens from the header.
Portrait videos keep captions, speed settings, and a time display.
PDF page and zoom controls support keyboard editing and show clearer feedback for invalid values.
Library search now includes chats.
Completed research stays selectable for video creation, and unprocessed recordings no longer appear to be generating.

September 5, 2026

Find, revise, and continue your AI work

Recently generated AI content now appears above Content on Home and in projects, including results in project folders.
Choose Create another version on a generated video to change its instructions, sources, or settings.
Your original video stays saved, and the new version uses your normal allowance.
Failed flashcard requests restore their reusable sources and creation settings, and video source failures can reopen for adjustment.
Closing recovery keeps the failed task available until you submit a replacement.

Schedule a Laureate downgrade

Laureate subscribers can switch to Premium at the next renewal from Billing settings.
Review the price and date before confirming, and undo the change before it starts.
Support staff can review, schedule, and undo a Laureate-to-Premium downgrade, with changes taking effect at renewal.

A steadier app under database load

Heavy background cleanup no longer slows your requests, so the workspace keeps loading and chat answers keep streaming.
If a database hiccup interrupts loading your workspace, the app reports it as temporary and retries instead of failing.

Smaller changes

Checkout failures stay visible with your billing period kept for retry, and plan buttons prevent overlapping checkout attempts.
Uploaded documents and audio keep their original names instead of getting automatic AI titles during processing.
Library filenames now keep spaces and non-English characters, and long recording names can be renamed without being cut short.

September 4, 2026

Large uploads and recording recovery

Laureate uploads can now use the full 1,000 MB allowance.
Oversized uploads return a clear size error instead of repeatedly retrying as a connection failure.
Recording pages tell a temporary loading error apart from a missing recording and offer a retry button.
When transcription finds no speech, the message asks you to check the saved audio before retrying or changing your microphone.

Work kept during AI recovery

Video retries keep their recovery model.
Long chat turns can continue beyond 50 tool rounds.
Flashcard decks keep completed cards when some source sections still fail after retrying.
A visible warning shows when a flashcard deck's coverage may be incomplete.

Smaller changes

Starting a transcription with an empty or unsupported file now shows the specific reason instead of an unexplained failure.
In a new project chat, the project picker now disappears as soon as sending begins, in step with the composer moving to the bottom.

September 3, 2026

Kimi K3, Muse Spark 1.3, and GLM 5.3 Flash

Moonshot AI's Kimi K3 and Meta's Muse Spark 1.3 are now Recommended in Chat and every AI creation tool.
Both can reason over images and long source material, with official provider logos in the model picker.
Kimi K3 and Claude Sonnet 5 use 2 AI Creation Credits, shown with a 2× badge, and Muse Spark 1.3 uses 1.
Chat remains unlimited on paid plans and never consumes creation credits.
Z.ai's GLM 5.3 Flash replaces Deepseek Flash, reads images and video, and saved Deepseek Flash choices move over automatically.

Combine more sources in one creation

Creations now allow 10 sources on Premium and 20 on Laureate (and their Teams equivalents), up from a flat 3.
Free creations use one source, matching the single Free upload.
Every creation tool, including Deep Research, now shares the same source limit.
The add-source picker now tells you which plan lifts the cap instead of only asking you to remove a file.

Uploads and creations recover from errors

Password-protected PDFs can now be unlocked and uploaded without repeating the same password error.
Dropped image, video-link, and recording requests retry safely without creating duplicate uploads.
Video Lectures recover once from a hard generation timeout before refunding and accept scripts slightly over length.
Flashcard decks that preserve every source question retry empty sections instead of discarding a finished bank.
Scholarly catches multi-audio selections before upload and replaces cryptic numeric errors with useful guidance.

Smaller changes

Chat and AI creation now address you as a user or learner, with guidance that covers professional and personal work too.
Personal Laureate is monthly only for new subscriptions; the $950 yearly option is gone, while Premium and Teams keep both.
Your purchase confirmation and chosen plan are now kept if you still need to finish sign-up after checkout.
Preparing a Laureate upgrade now retries through brief Stripe outages.
Google One Tap sign-in after tapping Get started on a marketing page now takes new accounts to onboarding, not Home.
YouTube imports now say when the transcript provider is out of capacity and let you retry later, instead of showing a rate limit.
Your task list and in-progress creations no longer drop out with temporary connection errors.

September 2, 2026

Gemini 3.8 Flash replaces Gemini 3.7 Flash

Gemini 3.8 Flash replaces Gemini 3.7 Flash in the paid model catalog, and saved 3.7 Flash selections switch over automatically.
Gemini 3.8 Flash is a standard one-credit creation model, while GPT 5.6 Sol and Claude Opus 5 remain the 3-credit Frontier options.
Flashcard Auto uses Gemini 3.8 Flash at low reasoning on Premium and medium reasoning on Laureate, and Free routing is unchanged.
Paid Video Lecture Auto now uses GPT 5.6 Luna with a higher reasoning budget, and Claude Sonnet 5 remains available to choose.

Creation modals and Home chat save into projects

Upload, Flashcards, Quiz, Video lecture, Podcast, and Deep Research modals now show a “Saving to” chip in the header.
The chip defaults to your most recent project, and you can switch projects, choose No project, or create a new project inline.
Sources and finished artifacts are saved together in the chosen project.
The Home composer now defaults to your most recent project, so answers come from that project's sources.
Pick No project on Home to keep global chat, and Home remembers your choice.

Pricing keeps your plan choice through sign up

Your Premium or Laureate choice on Pricing now carries through sign up, sign in, Google authentication, and onboarding.
The billing period you selected on Pricing is always preserved.
Your plan choice is kept even before the page finishes loading or when the button is opened in a new tab.
Premium plan screens now start on yearly billing, with monthly billing available in one click.

Smaller changes

Returning accounts can start creating on Home without waiting for optional shared and recent sections to load.
Finished recordings now go to transcription immediately instead of waiting on a separate duration check.
Video lectures, AI Slides, and AI Documents now retry automatically instead of failing when a processing workspace won't start.
Deleting your account now signs you out and confirms right away while data removal finishes in the background.
Fixed account deletion sometimes showing “Your account could not be deleted” for a deletion that was succeeding.
Fixed new accounts created with Google One Tap from a marketing page sometimes skipping onboarding.
Hovering a link to a stored file inside a page now shows the file name instead of a “Couldn’t read that link” error.

September 1, 2026

AI Slides and podcast slides get new layouts and colors

AI Slides now build from slide kinds like comparison, process, timeline, chart, equation, and quote, each with its own layout.
Slides draw diagrams, process flows, timelines, charts, equations, and icons from the same kit as Standard video lectures.
Decks are planned before they are written and checked for text overflow before rendering.
Podcast companion slides now pick a shape that fits each chapter, like a comparison, process, statistic, or timeline, not just bullets.
Both get a fresh color look per generation, and a "More options" panel lets you pick a specific look or switch to Dark Mode.

Faster team setup and a redesigned Team page

New teams follow one setup flow: name the workspace, choose model groups, invite teammates, set seats, then pick a plan.
Invite links and form choices stay in place when moving between setup steps.
The Team page now has separate Overview, People, Profiles, Settings, and Billing pages with icon navigation.
Invites live on People, model access and customization on Settings, and plans, seats, invoices, and receipts on Billing.

Recordings, YouTube videos, and infographics are kept safe

Finished recordings are now kept and processed even if measuring their length briefly fails.
Adding a YouTube video no longer fails when the transcript service is down, rate-limited, or slow.
Videos with no captions, deleted videos, and live streams still ask you to pick a different link.
Infographics and mind maps are delivered as soon as they render, even if search indexing hits an outage.

Smaller changes

Upgrade prompts now explain the specific Premium improvement for the action you tried before showing plans.
A YouTube transcript that is still being generated now shows as a wait you can retry, not a failure.
Follow-up emails about new uploads now wait for the file to finish processing so they can include study suggestions.
Picking a 0-byte file, such as an undownloaded iCloud or Drive placeholder, is now flagged immediately in every create flow.

August 31, 2026

Local pricing for Premium, Laureate, and Teams

Premium now shows and charges local prices in Japan, Saudi Arabia, Switzerland, New Zealand, the United Arab Emirates, and South Korea.
Premium local prices are also available in Hong Kong, Israel, Nepal, Argentina, Colombia, and Zambia.
Laureate now shows and charges the equivalent amount in supported local currencies.
Premium Teams and Laureate Teams now accept local currencies at checkout, with their global USD price unchanged.
Personal Laureate adds a $950 yearly option, saving about 20% compared with twelve monthly payments.

AI output and emails follow your language

With AI language set to Automatic, podcasts, mind maps, AI documents, essays, narrations, and research reports match your sources.
Exam feedback, generated PDF titles, folder organization suggestions, and text in chat-generated images also match your language.
Read-aloud now speaks Japanese documents with a natural Japanese voice.
Every email Scholarly sends, from sign-in to billing and team invites, now arrives in Japanese when your app language is Japanese.
Emails that are not yet translated arrive in English instead of failing to send.

Chat remembers earlier messages again

On most turns, the assistant had been seeing only your newest message and the reply just before it.
Chat now leaves room for conversation history, so follow-up questions and corrections build on earlier turns.
Details you established earlier in a thread now stay in view.

Smaller changes

Fixed an error that could stop an uploaded source from reaching the next creation step or make an upgrade button do nothing.
AI generations stopped by a provider's safety filter now continue with a different provider instead of failing.
Search now returns to fast, relevance-ranked results shortly after a deployment instead of staying on slower matching.
Marketing page titles no longer cut off at the “|” separator, so the full title shows in browser tabs and search results.
Japanese creation alerts now name the artifact you made, such as a mind map or infographic, instead of saying your PDF is ready.
When a pasted link can't be read, the message includes the specific reason again.

August 30, 2026

Scholarly in Japanese across the app

Choose Japanese in Appearance to use the Library, folders, Projects, and the main Chat experience in Japanese.
After you pick a tool on Home, the full source-to-create flow follows your app language, from upload to final settings.
Sign-up, account recovery, Teams, sharing, study modes, exams, editors, exports, billing, and errors are now in Japanese.
Names, file names, source content, and AI answers stay in their original language.
Signed-in URLs stay the same when you switch the interface language.

Deep Research updates

Team members can start Deep Research from Home's four main creation buttons, with Flashcards in the row below.
Deep Research now creates one PDF of the cited report by default instead of duplicate DOCX and Markdown files.
You still get separate files when you request them.
Research started inside a Project keeps that Project in the breadcrumb while the report is running.
The Thinking status animates its ellipsis, and the sidebar uses the same light-bulb icon as research reports.

Smaller changes

Fixed signup buttons on marketing pages that could leave visitors on the same page or send them back to Pricing.
Fixed shared links sending signed-out visitors to the login screen instead of the shared work.
Fixed video lectures failing when recovery fell back to a retired model; they now use your plan's current video model.
Fixed muted gray text in Billing history, weekly creation usage, and cancellation screens being hard to read in dark mode.

August 29, 2026

Deep Research rebuilt as a research workspace

Deep Research has its own sidebar destination, with workflows for literature review, market landscape, policy analysis, and due diligence.
Scope a run to a Project, attach internal sources, and choose the evidence mix, depth, model, and deliverables.
Finished work opens with a concise verdict, confidence and contested-claim signals, and open questions.
Read the cited report in Research, inspect sources in Evidence, and preview generated files in a full-screen workspace.
A dedicated chat beside the research already knows its reports, findings, and sources.

Core dashboard flows in Japanese

Choose Japanese in Appearance to use it on Login, Home, app navigation, the Home creation menu, and Connections.
Account, preference, support, billing, invoice, and usage settings are translated, along with shared AI language and voice controls.
App URLs stay the same, such as /home, /connections, and /settings, while /jp remains reserved for translated public pages.

Stronger AI Video Lectures

Video Lecture Auto now uses Claude Sonnet 5 with low reasoning for Premium and Laureate.
Free lectures keep their existing model, and a model you choose yourself is used exactly as selected.
Illustrated lectures add labels after the artwork is generated, so instructional text is spelled exactly.
Standard lectures check every requested topic individually and pace longer explanations through more meaningful visual beats.

Smaller changes

Add Content now shows the upload slots your plan has left, with Library and Drive still available.
Home's Content count now includes items that are processing or ready to open.
Free's automatic model is now labeled Auto (Free).
Lifetime Billing allowances no longer suggest a daily reset, and limit banners now focus on the next action you can take.
Podcast Deep dive badges now count the notes actually shown, and the outdated Short Videos 'New' badge is removed.
Explore search results no longer mix in unrelated recent work.
Explore cards show the matching excerpt when a result matches inside a page or PDF, and garbled older titles are repaired.
True-or-false quiz prompts no longer appear as answers on the flashcard front.
Exam feedback no longer adds reaction emoji after mistakes.
Chat's exact-source rules now protect adjacent repeated digits in multilingual identifiers.
Chat replies use a lighter body weight for easier reading, while headings, bold, and other emphasis are unchanged.
If chat's file workspace can't start, the assistant now says so once and answers with what it has instead of retrying in a loop.
Study guides, slide decks, and other generated documents no longer fail when the connection drops briefly while loading sources.

August 28, 2026

Chat can set up projects and scheduled work

When a conversation turns into ongoing work, chat offers to create a project and move the chat into it.
Everything the chat produces after that is filed in the project.
Chat can find your projects and past chats by name and look inside a project.
Chat can rename, file, or put away your pages, files, folders, and projects on request.
Name a time, like an exam on Friday, and chat offers to have a study guide, deck, or summary ready then, shown on Home.

Chat with a project from anywhere

The chat box on Home and in full-screen Chat has a project picker that grounds your next question in that project's sources.
Picking a project brings along a chat you already started, and switching projects takes one click.
A new expand button on the project page opens the conversation full width on the Chat screen.
Creating a project from the chat picker now asks for its name before creating and selecting it.
After a conversation starts, the extra project picker is hidden because the breadcrumb shows the chat's project.

Connect only the Google Drive files you choose

Google Drive has its own Connections page with per-account approval, default, reconnect, and disconnect controls.
A Drive connection covers only files you select through Google's Picker, including Docs and Slides.
Chat can search and read those files but cannot browse the rest of your Drive or edit your files.

Smaller changes

Chat answers are shorter and lead with the answer, with less preamble, fewer headings, and no closing recap.
Form fields no longer grow a border, ring, shadow, or scale effect when focused.
Cmd/Ctrl+K search now finds project chats and chats over PDFs, pages, recordings, and other items, matching on what was said.
Home uses larger tool names, keeps Start Recording in the main row, and moves Text-to-Speech and Transcribe into More tools.
AI Mind Map, Worksheet, and AI Infographic are now document types inside AI Document Creator.
Changing the dashboard language now keeps you on the same address, such as /home or /chat.
Chat now waits out a brief network outage when staging your file instead of giving up and stalling the reply.
Fixed finished PDF study guides sometimes showing as still working or failed.
Emails about a new PDF upload now wait for the file to be read, so they include its questions and key points.
Uploading a hidden "~$" Office lock file now says so, instead of suggesting your document is corrupted or password-protected.
Billing history now loads even if one document link is malformed, and shows a spinner instead of gray placeholder rows.
A shared project's collaborator list now renders when an access request has no readable email address.
A settings address with no page behind it no longer opens an empty panel.

August 27, 2026

Explore rebuilt with a design for each format

Short Videos are vertical, and Podcasts show episode artwork and waveforms.
Flashcards show a real question, and Research shows its abstract and sources.
Generated presentations, study guides, and pages each have a format-specific design.
Every discoverable Explore category is now available in English and Japanese, with clearer search.
Privacy-sensitive formats such as meeting notes and folders stay out of discovery.

Shared projects become real workspaces

Invited people and teammates with access can open shared projects from their own Projects list.
Collaborators can chat across the shared sources, edit the shared brief and instructions, and see other active collaborators.
Collaborators get a read-only view of all files, pages, folders, and the items inside those folders.
Each collaborator keeps their own Private Chats and Private Memories, and project chats are never shared.
Only the owner can add or remove content, manage members, change the default profile, or archive the project.

Guided team setup

Creating a team now follows four steps: name your team, invite teammates, pick a plan, then choose seats.
Invite teammates by email or with a shareable join link.
Seats are pre-filled to cover everyone you invited.
Invites beyond the workspace's open seats are sent automatically as soon as your plan activates.

Smaller changes

Until October 19, 2026, workspaces without a team plan included 3 seats; new workspaces now start with only your own seat.
Members and admins can now leave a team from the Team page's Overview tab, which frees their seat right away.
The flashcard editor now has Undo and Redo controls, with Cmd/Ctrl+Z and Cmd/Ctrl+Shift+Z shortcuts.
Practice no longer drops signed numbers or percentages from true-or-false statements before grading.
Generated Slides now show multilingual text and punctuation correctly instead of encoding garbage.
Video Lecture companion PDFs now include searchable, selectable text for assistive technology.
Keyboard focus now stays inside Settings while it is open, and the rest of the app is hidden from assistive technology.
On Pricing, switching to Business now compares Premium for Teams and Laureate for Teams, including per-member allowances.
Settings has more room and groups options into Account, Workspace, AI, Billing, and Support sections.
Billing now offers downloadable invoice PDFs and Stripe receipts, plus team billing documents for owners and admins.
Fixed Podcasts, AI Meeting Notes recordings, voice previews, and video not playing when Scholarly opens inside another app.
Chat now says a YouTube transcript is still processing instead of failing as if the video were unavailable.
Seeking while a recording is paused now moves the progress bar and highlights the right topic immediately.
AI Document, Essay, and Spreadsheet creators now clear earlier errors before showing upgrade options at your AI limit.

August 26, 2026

Transcribe audio, video, and YouTube into a file

The new Transcribe creator turns up to 60 minutes of audio, video, or a YouTube video into a file.
Choose a clean .txt file or timestamped .vtt captions.
Each transcription uses one AI Creation Credit for the whole source and runs in the background.
If a transcription fails, the credit is returned automatically.
The narration feature is now named Text-to-Speech everywhere.

Project chat can edit and organize your project

Ask project chat to rewrite a page, update a flashcard deck, or edit a file anywhere in your project.
Every edit arrives as an approval card: accept to apply it or deny to dismiss it, and nothing changes until you accept.
Approval cards keep working even if you come back to them later.
Chat can also rename, archive, restore, or move pages, files, and folders, and create new folders.
Archiving is always reversible, and anything chat creates is filed into your project, even if you close the tab.

Separate app and AI language settings

Appearance now includes an App language control for Scholarly's interface.
The existing output-language control is now named AI Language under Chat.
Changing the language of menus and navigation never changes the language of AI responses or new creations.

Smaller changes

Deep Research no longer requires an extra file and asks what to produce, such as a cited report, slides, or a study guide.
Flashcard decks with duplicate or missing card identifiers are repaired automatically, so Practice and Exams no longer fail.
Chat sidebars for content with existing conversations now lead with recent chats instead of the full Chat introduction.
Free accounts that have used their upload now see the plan window when starting AI Meeting Notes, not after recording.
A recording that has already captured audio always finishes, even if the allowance runs out while it is running.
When the free quiz allowance runs out mid-session, the plan window now opens instead of a warning while answers go unsaved.
YouTube videos whose transcript is still generating are added to your library right away instead of failing after two minutes.
Fixed pasted text losing its colors when they arrived in a CSS format rather than a hex code.
Fixed opening a second PDF preview still showing the first document until the page was reloaded.
Word, PowerPoint, and Excel files with stray spacing in their names now import instead of being rejected.

August 25, 2026

Projects, reports, and files reopen instantly

Returning to a project, research report, recording, or file shows what was already on screen instead of a loading spinner.
Moving between two projects is immediate after the first visit.
Reload the page whenever you want to pull in changes made elsewhere.
Opening a research report by its link now shows a loading indicator instead of briefly showing “Research session not found”.

Better Slides layouts and accurate Podcast details

Generated Slides use the full canvas more consistently.
Tables in Slides stay in readable rows and columns.
Each slide keeps its own styling in the final PDF.
Podcast pages show the speakers and voices actually present in the transcript, including solo and larger casts.

Smaller changes

Fixed requesting a sign-in link from the “Email sign-in link” page failing with an “Email is required” error.
Removed Weekly Brief from Home; Scholarly no longer creates or emails automatic briefs from recent material.
When Premium's 10 weekly AI Creation Credits run out, the first button is now Keep creating (80 this week); credit packs stay secondary.
The prorated amount due today and 20% off the next two full Laureate months now appear in the plan window, not on the credit wall.
Older true-or-false questions saved with the wrong answer key are repaired before Practice or Exams use them.
Chat, Slides, and Video Lectures keep multilingual identifiers, signs, percentages, decimal precision, and scientific notation exact.
Fixed completed Slides sometimes disappearing after they finished generating in the background.
Story Books can use your Custom Designs on every plan, as intended.
YouTube summaries can be retried while captions are still processing.
Flashcards now guide you to replace an image source that is no longer available.
Short Video is now described as a short video format rather than a fixed-length promise.
The Add content window now shows when all three source slots are full, disables extra pickers, and asks you to remove one first.
Fixed the page editor crashing when pasting part of a table with row-spanning cells.
Fixed shared conversations and Explore category pages briefly flashing a “not found” message while loading.
Pressing Finish on a recording that was already processing or ready now takes you to it instead of reporting a failure.
The AI Video Lecture preview shown before upgrading displays its cover image again.

August 24, 2026

Google Drive files in Chat with Connections

Connections now has its own place in the sidebar.
Link more than one Google account and Chat can search and read the Drive files you share with Scholarly.
Chat answers are grounded in your own documents.
Access is read-only and per-file: Scholarly sees only what you share through the Drive picker and cannot change or delete anything.
Each account is managed independently, with one set as the default for Chat.

Chat switches models when a provider disconnects

If Grok or another AI provider closes a response before any answer reaches you, Chat moves to a compatible backup model.
You no longer get a generic error in that case.
If part of the answer or an action has already started, Chat stops safely and lets you retry.
Chat does not mix two models in one answer or repeat an action.

Scheduled tasks start at your local time

Chat now uses your current time zone for relative and unqualified times such as "2 AM my time".
Future tasks are labeled Scheduled everywhere they appear.
Scheduled tasks cannot be picked as sources until they have finished.

Smaller changes

Deleting your account now recognizes a deletion in progress and signs you out instead of showing "Failed to delete account".
Deleted, private, or region-blocked YouTube videos now show as no longer available, with a suggestion to try a different link.
The assistant keeps the conversation going when a YouTube video is unavailable instead of stopping on the error.
Text-to-Speech narrations are no longer offered as sources in Chat, Flashcards, Quiz, Podcast, Research and other create flows.
A narration attached before this change now shows that it cannot be read, with a suggestion to attach the source document.
Narrating a document again with a different voice, speed, page range or title now counts as a separate narration.
Repeating an older essay or narration in a later week now draws on the current week's allowance.
Asking for a premium video quality no longer opens the "Free AI Creation Credits used" wall if you have not spent a creation.
Going over a plan's PDF page limit now opens the right upgrade screen instead of an upload or model message.
The Laureate card in Billing settings now shows your local price instead of always quoting $99.
Going back a question in practice no longer clears the answer on screen when the previous question is no longer in your set.
Scheduling a task no longer lets you confirm a time that passed while the picker was open.
A scheduled Study Pack now tells you when one of its files could not be queued.
Downloading an image from the Image Studio now works reliably in Safari and Firefox.
The Frontier models help panel is back in the chat model picker.
The Laureate and flagship credits Help Center links now open the article instead of a Not Found page.
Slides, PDF and other document creations that are still running are no longer stopped, marked failed, and refunded.
Canceling an infographic or mind map just as it finished or failed no longer overrides the result or keeps the AI Creation Credit.
Downloads through the file proxy now report an unreachable remote server instead of hanging.
Text-to-Speech no longer fails outright when a language name is not recognized.
The Flashcards file picker no longer offers spreadsheet files it would then reject.

August 23, 2026

Schedule AI creations for later

Schedule any one-time AI creation for a specific day and time instead of starting it immediately.
Scheduled tasks from creation windows or Chat stay together on Home.
Tasks created inside a project also appear in that project's Tasks view.
See exactly when each task will begin, or choose Run now to start it early.

Images are now first-class files

Upload JPG, PNG, GIF, WebP, BMP, TIFF, HEIC, HEIF, and AVIF images anywhere you add sources.
Add images from Home, Chat, folders, projects, and creation tools.
Open images in the file viewer, rename, move, download, archive, or chat with them like your other files.
Use images for Flashcards, Practice Quizzes, Podcasts, and Video Lectures.
Images placed inside a page remain editable image components.

Long podcasts, recordings, and videos stream reliably

Files over about 32 MB, which includes most full-length podcasts, no longer stop with an error during playback.
Long files now stream in parts, so playback and seeking work the whole way through.
If a file still won't load, the message explains it is a temporary delivery problem and suggests trying again or refreshing.
Players no longer suggest uploading or creating your audio or video again when the file is fine.

Smaller changes

Video lectures drop Kids and 480p: resolution is now 720p or 1080p, with Dark Mode, Look, and background music under More options.
Standard video lectures always include the companion deck, and Podcasts no longer offer Quick Summary.
Expert styles are removed from smaller tools such as lesson plans, cheat sheets, PDFs, essays, timelines, worksheets, and mind maps.
Infographics and mind maps no longer offer Freestyle (pick a named look instead), while Story Books keep every art and expert style.
Opening scholarly.so while signed in now goes straight to Home, and the public landing page no longer disappears while you wait.
Deep Research from chat and other background creations now keep their real titles in project lists and completion emails.

August 22, 2026

Text-to-Speech: listen to any document as one MP3

A new Text-to-Speech tool reads a whole document aloud and saves it as a single MP3 in your library.
Start it from More tools on Home, or from the creation actions on any file, page, or PDF.
Choose from over 25 voices and a narration speed between 0.75× and 1.5×.
It costs 1 AI Creation Credit per 10,000 words, up to 30,000 words per narration, and the cost is quoted before you start.
It runs in the background, opens the audio when ready, and strips formatting characters so the voice reads prose.

A cleaner, quieter interface

Buttons, inputs, and panels now sit on subtle background steps instead of boxes and outlines.
Color is now used only for actions and alerts that need it.
Plan limits for weekly creations, uploads, and chat now look and read the same everywhere.
Sharing now uses a single link, and the share dialog no longer shifts as it loads.

Slides use the whole canvas

Generated Slides now plan a visual story before drawing and vary layouts across the deck.
Slides with too much empty space are now treated as unfinished.
Tables now render as real horizontal tables with readable headers and columns.
Long tables now split cleanly across slides.

Smaller changes

Home, the Library, and Projects keep what was on screen when you return, including Library filters, search, and scrolled rows.
On Premium, reaching the weekly AI Creation Credit limit now shows when credits renew and what Laureate adds (80 a week, 1 GB uploads).
The Laureate card now shows the prorated amount due today and the monthly price after that.
The credit packs window now shows this month's spend and per-creation cost next to what those creations would cost on a plan.
Chat, Slides, and Video Lectures now keep signs, percentages, multilingual text, and identifiers exactly as in your source.
Exam answer keys now accept answers that match the source's wording, and medium Flashcard decks stay proportional to shorter sources.
Finished Slides and other documents stay available from their task while being added to Home, instead of briefly disappearing.
New Podcast and Slides pages now load consistently, and completed quiz results are titled “Quiz results”.
On the Free plan, choosing the standard 2-minute length for podcasts and video lectures no longer blocks the creation.
Upgrade prompts for paid creations now open the plan screen with the matching feature highlighted.
The Home prompt box now has the same 5,000-character limit as the create panel, so long prompts no longer fail the creation.
Some new accounts pick AI agents after naming their first project. Files stay on the device, and creation starts after checkout.

August 21, 2026

Customization Profiles for every AI creation

Customization Profiles replace Saved Instructions and Custom Designs, combining instructions, a preferred AI model, and visual references.
Visual references can include logos, brand guides, or templates, and one-off instructions and format-specific themes stay available.
Choose a profile from the new button at the top of every Customize step.
Existing saved instructions and designs are moved into Personal profiles automatically.
Team admins can publish shared Team profiles from the new Profiles tab, and each Project can choose a default profile.

One credit balance for AI models

Frontier Boosts have been replaced by model-based AI creation costs.
Standard models use 1 credit, while Frontier models GPT 5.6 Sol and Claude Opus 5 use 3.
Every paid plan can choose from the full creation-model catalog.
GPT 5.6 Sol and Claude Opus 5 are no longer offered in unlimited Chat.

Personal and Business plans

Plan options now open with Personal and Business tabs, so you can price a plan for yourself or per seat for your team.
Being in a team no longer locks you out of a personal plan, and you no longer need a team to see Business pricing.
Yearly and Monthly now sit inside each plan card, and the term you pick on /pricing is carried into checkout.
Team setup is now a full-screen, three-step flow (name, plan, seats) showing what each seat gets, seat pricing, and a running total.
/pricing uses the same plan cards as the app and shows Premium for Teams and Laureate for Teams side by side.

Smaller changes

Chat answers keep running if you close the tab or leave the page, and reopening the conversation reconnects to the same answer.
Recent chats and Chat results in quick search now show the AI model used in that conversation, with its provider icon.
The Teams tab on /pricing is now called Business, and the separate educators box has been removed.
Fixed choosing a team plan sending you back to the last onboarding slide; it now takes you to team setup.
Guided first-project setup (for some new accounts) now accepts documents, spreadsheets, text files, images, audio, and video.
Flashcards now name any file that can't be read as text or an image instead of silently leaving it out of the deck.

August 20, 2026

A guided first project for new accounts

Some new accounts now start by naming a project, saying what it is for, and adding a file, URL, or pasted text.
Recommendations match your goal, from certification and work material to research, teaching, and coursework.
Study guides, practice sets, slides, flashcards, videos, podcasts, and mind maps are all available to create.
A new setup step tunes style, length, deck size, resolution, and AI model before a creation starts.
Free projects build one creation at a time; paid plans can pick several and unlock longer, higher-quality settings.

Lower Teams pricing and team upgrades

Premium for Teams is now $35 per seat each month or $336 per seat each year.
Laureate for Teams is now $110 per seat each month or $1,056 per seat each year.
Annual billing costs 20% less than paying monthly for twelve months.
In a Team workspace, upgrade buttons now say Upgrade Team and open per-seat Team pricing.
Owners choose Premium or Laureate and set seats for the whole team; members are directed to whoever manages billing.

Smaller changes

Files that are too large or over your upload allowance are now named right away, with the upgrade that lifts the limit.
Fixed team and workspace checkout failing with a repeated server error instead of opening Stripe.
Skipping the first Team plan choice opens Home with the Team welcome, and the duplicate Choose plan button is gone from the Team page.
Canceling a Team checkout now returns Home without the personal checkout offer.
Saved web pages built with scripts now keep their notes instead of being rejected as not enough study material.
Saved pages that hold only the page shell are now flagged right away, with what to add instead, before a creation is spent.
Fixed companion slide decks failing to build after a finished video, which restarted the whole generation.
Long Chat research turns now write up their findings instead of ending empty when they near the step limit.
Chat on videos, PDFs, notes, projects, Image Studio, and other sources now says it already has that source in front of it.
New accounts start with the chat rail open; Scholarly keeps your choice if you have shown or hidden it before.
Fixed two-minute video lectures failing when a narration scene ran a few seconds longer than estimated.
Plan options now list length limits: Free is 2 minutes; paid plans allow 60-minute podcasts and 45-minute video lectures.

August 19, 2026

Upgrade badges in the model picker

Models that need a paid plan now show an Upgrade badge in Chat and in every creation tool.
Choosing one of these models opens your plan options instead of failing when you send.
Recommended models no longer look available and then get rejected.

One email when any creation is ready

Settings → Notifications now has a single “When a creation is ready” toggle.
Scholarly emails you when any long-running creation finishes, not only podcasts, videos, and flashcards.
This includes video lectures, study guides, PDFs, slides, research, and more.
Uploaded PDF processing and meeting-notes transcription still do not send an email.

August 18, 2026

One Home input that runs every tool

Pick Chat, Image, Document, Deep Research or Essay on Home's input and describe what you want.
Pick Study Guide, Slides, Mind Map or Worksheet and drop a PDF, doc or deck straight onto the input.
What you type or drop carries into that tool's setup, so you are not asked for it twice.
Chat on Home is the real chat input, with linked sources, pasting, attachments, the model picker and voice.
Every other tool sits under More tools.

Record anywhere, with a live transcript

A floating bar follows you across the app with elapsed time, a live level meter, and pause and finish.
Leaving the recording page no longer stops capture.
AI Meeting Notes stream the transcript as you speak, with a status showing listening, transcribing, or hearing nothing.
The recording page leads with title, status, and date, then one card for Transcript, Summary, Action items, and AI Creations.
Recording warnings and errors now appear as one persistent notice with the fix or retry attached.

Generations finish even if you leave the page

Study guides, slides, worksheets, reports and other documents keep generating after you leave the page.
An interrupted generation picks back up with your sources in place instead of starting over.
If a generation fails, you get a real reason and your credit back.

Smaller changes

Recommended Models now appear on the model picker's main menu under Auto, and Auto rows no longer show the underlying model.
Retrying a video lecture made with a since-replaced model now re-runs it on your plan's current model.
Video lecture narration now retries through brief speech service outages and explains the error if it still fails.
Fixed linked sources showing as an empty gray bubble until reload; they now appear as a labeled chip when you send.
Flashcards now tell you which sources were left out of a deck, such as a PDF still processing or a video still transcribing.
Rename a project or AI Meeting Notes by clicking its title, and rename folders from “Update Name” in All Content.
Live transcripts of very long recordings no longer drop their earliest lines.
Fixed true/false exam items marking correct restatements false, and mind map labels shrinking to unreadable sizes.
Slide decks no longer leave the bottom quarter of the page empty, and video lectures now respect the length you picked.
Recordings in Spanish, French, or German now get their summary, topics, and title in that language instead of English.
Long recordings are no longer mistaken for empty ones, and stretches with nothing captured are marked instead of filled in.
Interrupted recordings now expire on their own instead of blocking new ones, and saved audio stays available to process.
Documents quoting text in Hindi, Arabic, Russian, or Chinese now render it properly instead of dropping or translating it.
Checkpoints and chapters on long videos now spread across the whole runtime, and checkpoints are no longer silently discarded.
Added Japan-specific exam pages for 宅建, 日商簿記, ITパスポート / 基本情報, 医師国家試験, TOEIC, and 社労士.
Reports, timelines, SOPs, spreadsheets, and more now start from one AI Document page; older tool pages send you there.
Added new pages for reports, case studies, and rubrics.

August 17, 2026

Buy AI Creation Credit packs

Paid plans can now buy one-time AI Creation Credit packs: 5 for $15, 15 for $30, or 40 for $60.
Credits never expire.
Credits are used automatically once your weekly allowance runs out.
Buy packs in Billing settings or right where you hit the limit.

AI Meeting Notes survive sleep and mic dropouts

Live capture now detects when your computer sleeps, the tab is suspended, or the microphone dies mid-recording.
The mic restarts automatically where possible, and you get a warning on the spot.
Finished notes clearly show how much time went uncaptured instead of quietly delivering a partial recording.

Smaller changes

Free AI Chat now includes 3 messages total, account-wide with no reset; Premium includes unlimited AI Chat.
Hitting the free AI Chat allowance mid-send now shows the upgrade path instead of a generic error with a Retry button.
Removed the Up next card from the top of Home; Weekly Brief still appears when available.
Retrying a failed video lecture now restores full automatic recovery, so it gets a real second attempt.
Scholarly Laureate ($99/month) now shows in your local currency on the plan modal and pricing pages, matching checkout.

August 16, 2026

Start at 2 minutes, then make it longer

Free podcasts and video lectures now use a fixed 2-minute length.
A persistent Make Longer banner can upgrade and replace either item in the same spot.
Making an item longer does not use another creation credit.
The original stays available until the longer version is ready.

A redesigned Weekly Brief on Home

Weekly Brief now uses the same calm, neutral design as the rest of Home, with a clearer reading hierarchy.
The Ask AI prompt is more focused and the self-check is cleaner.
Adding material for next week is simpler.
Your first Weekly Brief can be created after one full day as a subscriber instead of five, if you have recent material.

Stop active creations from Chat

Ask Chat to stop an active podcast or AI video lecture.
Scholarly cancels the running work and does not leave it running in the background.
The stopped item is marked as canceled everywhere.

Smaller changes

Premium members billed in a local currency can now upgrade to Scholarly Laureate in-app at $99/month, converted at current exchange rates.
PDF imports now keep Arabic and Japanese identifiers in logical reading order instead of reversing digits or spacing characters apart.
Flashcards now read diagrams, labels, and other visible page evidence alongside selectable PDF text.
Fixed question-bank flashcard decks failing when part of the source, like a video's outro, contains no questions.
Video Lecture narration now recovers automatically from brief audio-service connection drops instead of failing.
AI video lectures now include fonts for Thai, Hindi, Arabic, Chinese, Japanese, Korean, and other scripts, so text no longer shows as boxes.
Fixed Short and Kids & Story video lectures failing to finish when the narration contained certain words.
Chat now correctly explains that AI Meeting Notes use the file-upload allowance rather than an AI Creation Credit.
Home no longer offers processing or failed items as ready to open.
Flipping through Slides quickly no longer blanks a page that is still loading.
Mobile timestamps are easier to read in light mode.
The yearly billing badge now stays inside its option card on the exit offer.

August 15, 2026

One AI Document creator

PDFs, reports, timelines, SOPs, spreadsheets, tables, exports, calendars, and more now start from one AI Document tool.
Pick your sources in a single source picker, then choose the document type.
Generated PDFs now display 13 languages correctly in every theme, including Arabic, Chinese, Hindi, Thai, Hebrew, Russian, and Dzongkha.

Set up a team workspace at registration

Registration again asks whether you're starting personally or with a team.
Choose Team to name the workspace, then set the number of seats and choose Premium or Laureate.
The same team setup is also available later.
You can try the owner workspace with your personal plan before paying, and invitations and join links unlock after checkout.
Returning to checkout later keeps the same workspace and content.

Longer, more accurate paid video lectures

Paid video lectures can now run up to 45 minutes at every resolution.
A chosen length now delivers a lecture that fills that time instead of a short recap.
A selected brand kit now applies to the default Standard lecture.
Watermark-free lectures come from generating with the Scholarly logo turned off, and free lectures stay branded.
Math video mode now starts with source-grounded animated scenes when one fits, and uses the existing math-video style otherwise.

Smaller changes

Practice questions and exams now check true/false answer keys against exact flashcard evidence, including numbers, signs, and symbols.
Flashcards, Questions, and Exams now preserve scientific operators, zeroes, unit separators, subscripts, and superscripts.
Video and podcast captions no longer split decimals or units, and short video captions keep identifiers like ORION-P9 exact.
AI Slides, Standard Videos, and Podcast companion slides no longer cut off text, units, or source figures.
AI Chat now asks for your explicit approval before sharing product feedback with the Scholarly team.
The welcome tour and exam results now describe Free chat, quiz, voice, and exam allowances as lifetime limits that do not reset.
When more than one Free allowance is reached, Home now names exactly which allowances were used.
Registration can be submitted with Enter, and Settings billing links now open the Billing tab.
The Home composer is easier to see in light mode, and plan comparisons fit more comfortably on screen.
The post-purchase welcome no longer asks you to change billing periods, and monthly members no longer see a yearly-billing popup.
The Flashcard upload helper now lists PowerPoint among accepted presentation files.
Repair Card, whole-deck Refine with AI, and card edits in deck chat no longer use an AI Creation Credit.
Card repairs and whole-deck refinements now use medium reasoning for more reliable corrections.
Empty question-bank flashcard decks are now automatically retried on a different model instead of failing.
Home now waits for your workspace to load, so new accounts no longer see the wrong screen flash before getting started.

August 14, 2026

More room to try Free, with clearer limits

Free AI Chat now includes 10 messages total instead of 3 per day, and the allowance follows your account without resetting.
Free autocomplete, quiz practice, and voice mode now have lifetime allowances of 500 words, 5 questions, and 15 minutes.
Free practice exams include 1 completed exam attempt, and only completed attempts count.
Settings, upgrade screens, pricing, and Help now show lifetime usage consistently.
A failed Chat response now gives back the message it reserved.

Image Studio is out of beta

The Image Studio no longer has a separate daily beta cap.
Each new image or reference-image edit uses one AI Creation Credit per selected model.
Failed generations are refunded automatically.
Images created or edited in regular AI Chat do not use AI Creation Credits, and the normal Chat message allowance still applies.

Stronger, more forgiving flashcard creation

Premium Auto flashcards now use Gemini 3.7 Flash with medium reasoning, and Laureate uses high reasoning.
Free Auto flashcards stay on GPT 5.6 Luna.
A completed deck with fewer than 15 cards no longer uses an AI Creation Credit, and Scholarly tells you when it was returned.
Upgrading from Add content now restores the files, pasted text, links, Library items, and Google Drive sources you selected.

Smaller changes

Long YouTube videos without captions no longer stall before failing, and retrying picks up the same transcript job without duplicating it.
YouTube imports now ride out brief transcript-provider hiccups, and video thumbnails degrade gracefully when generation is unavailable.
Dragging a video file onto a page's video block now uploads it instead of doing nothing.
Switching between the Upload and YouTube Link tabs in a video block no longer clears your link or chosen file.
Fixed the featured lecture title being cut off on Explore and the podcast and video charts running off the screen on phones.
Comparison tables in blog posts now scroll sideways on a phone instead of losing their last columns.
File preview now opens edge to edge on a phone, with PDFs fitted to screen width and the file name no longer squeezed out.
Dragging the podcast and recording progress bars on a phone now scrubs instead of jumping to a chapter.
Model pickers and other menus opened near the top or bottom of the screen now flip or slide into view.
Notifications that offer an action, like opening a finished deck, now wait for you instead of disappearing after a couple of seconds.
Focus mode results no longer cut off the score and buttons, and long exam answer choices show in full on a phone.
On a phone, the Image Studio opens on your images with the chat a tap away, instead of the chat covering the screen.
The page export dialog stacks its options and preview on a phone and keeps Export within reach.
Plan limits and upgrade prompts stack properly on a phone, and pricing table feature explanations can be tapped open on touch screens.

August 13, 2026

Image Studio chat and Home shortcut

Image Studio now appears in Home's Create with AI tools, so you can start without opening More first.
A new Image Studio chat turns natural-language requests into background image jobs.
Model, aspect ratio, and quality controls set defaults that follow-up instructions can override.

Gemini 3.7 Flash

Gemini 3.7 Flash replaces Gemini 3.6 Flash in Chat, Research, Slides, Video Lectures, and Infographics.
It is Google's latest Flash model for chat, research, and visual work.
Saved 3.6 Flash picks return to Auto, and you can choose 3.7 Flash from the picker.

Laureate upgrades that finish in Scholarly

Premium members can review the exact prorated amount and confirm a Laureate upgrade without the billing portal.
If Premium was scheduled to cancel, confirming Laureate reactivates the subscription as part of the same upgrade.

Smaller changes

Long recordings that failed with "something went wrong" when stopped now finish and save.
Adding a YouTube link for a live stream now explains that captions only appear once the stream ends.
The chat button on PDF, presentation, recording, podcast, video, and other content pages now reliably opens the assistant.
Pressing Send after reaching your Chat limit now opens the plan comparison instead of appearing to do nothing.
Chat limits now explain when they reset without interrupting your draft.
Chat, Podcasts, and Video Lectures now keep requested source identifiers exact, including multilingual text, spaces, and punctuation.
When a file is too large, Add content now names your plan's limit, the next plan's limit, and what to try instead.
The "Add more files" area now shows the size limit before you pick a file.
Research follow-up reports now start with your uploads and earlier report files, and Research Chat can inspect them directly.
On the free plan, Download is now the main save button and opens upgrade, while copying a share link stays free.
Every AI creator now allows 5,000 characters of custom instructions, and selected presets and designs always show a checkmark.
Podcast and video tabs keep full labels on smaller screens, and Home tool labels no longer clip at larger zoom or scaling.
Two-speaker podcast requests now stay limited to the host and guest through transcript and audio generation.
The AI Video Lecture creator now opens immediately instead of showing a dark loading pause on first launch.
Chat and AI creation no longer show a Frontier badge on every Frontier model; they remain under Frontier Models.
Cards in a batch that came back without practice questions are now regenerated automatically, so none are blank in study mode.
Every creation dialog now checks pasted websites right away and keeps blank pages, loading screens, and sign-in walls out of sources.
Flashcards now retry once on another compatible AI model when a readable page is wrongly dismissed as blank or front matter.
PDFs, images, spreadsheets, and other files created in chat now open in the side preview instead of failing or leaving the chat.
Chat now previews notes and files saved without a full file type, and duplicated videos and flashcard images still play.
Downloaded PDFs keep a .pdf suffix even when the title has no extension, and file page downloads keep the right extension.

August 12, 2026

A rebuilt pricing page

Plans and prices now sit at the top of the pricing page, and each plan says who it is for in one line.
The plan comparison table now fits on screen instead of scrolling sideways and cutting off the Laureate column.
Teams now offers Premium for Teams ($45/$324) and Laureate for Teams ($135/$972) under a Personal / Teams switch.
The monthly/yearly toggle is gone: Premium is yearly-only and shows the monthly price instead of the yearly total.
Laureate creation Frontier Boosts doubled from 10 to 20 each week, with 75 weekly AI Creation Credits unchanged.

Grok 4.6 and more models for Premium

Grok 4.6 from xAI replaces Grok 4.5 across chat and create tools, as a recommended Frontier model with thinking always on.
Previous Grok 4.5 and Grok 4.3 selections continue automatically on Grok 4.6.
Premium now includes recommended access to Grok 4.6, Claude Sonnet 5, Gemini 3.1 Pro Preview, and GPT 5.6 Terra.
The full supported catalog is back under More Models, and flashcards keep GPT 5.6 Luna as their focused recommendation.
Scholarly remembers your model choice for each AI creation tool, including Podcasts, Video Lectures, Research, and Slides.

Longer free podcasts and video lectures

Free podcasts now run up to about 15 minutes, and upgrading still unlocks a longer target length.
Free AI Video Lectures now use Auto length, sized to your sources (typically a few minutes), instead of a fixed 2 minutes.
Paid plans can pick a specific video lecture length up to 45 minutes.
Podcast joins Flashcards and Video Lecture on Home, after upload, in the New menu, on source AI actions, and as a first-run welcome choice.

Smaller changes

Feature, use-case, and landing pages now open with a real product screenshot beside the headline, including pages that had no imagery.
Feature and use-case pages now share the same headings, cards, and spacing for a consistent look.
Image Studio and shared xAI image tools now use Grok Imagine Image 2.0 for image generation and reference-image edits.
Laureate now highlights the full Frontier model catalog (Claude Opus 5, Gemini 3.1 Pro, Grok 4.6) on 20 creations and 20 chats each week.
Auto (Laureate) uses Claude Opus 5 at low reasoning for Frontier work, then GPT 5.6 Luna after weekly Frontier uses are gone.
Standard video lectures can now use a color look or a saved custom design, and companion slide decks are included automatically.
Billing now shows Laureate alongside Premium and opens its $99/month checkout path.
Plan details now separate unlimited standard-model Chat from 20 weekly Frontier Chat boosts.
Stripe checkout summaries now list concrete Premium and Laureate allowances.
YouTube videos that need transcribing are now processed once and reused, instead of sometimes failing or being transcribed again.
Folders on Home now appear above Content.
Removed the community banner and Teams contact card from Home. Teams contact is in pricing and billing.
Word and PDF page controls now move immediately and reliably, including at the last page of a short document.
Video lecture companion deck downloads are now labeled as a slide deck instead of a video.
AI Slides now render repeated formulas identically, keeping signs, decimals, superscripts, and subscripts consistent across the deck.
The homepage feature tour no longer repeats Flashcards, and user counts across the app now consistently say 170,000+ users.
Chat keeps source numbers exact on request, including leading plus signs, trailing zeros, precision, units, and inequality symbols.

August 11, 2026

Research workspace with built-in chat

Research is now a focused workspace with separate tabs for the report, its sources, and creations, plus a dedicated chat.
Use the chat to summarize, compare findings, challenge a conclusion, or inspect the evidence without leaving the report.
Request follow-up research from the chat to start another background report in the same Research session.
Deep Research from Home or an old Research link opens a guided setup for sources, objective, depth, evidence priorities, and model.
Choose at least one extra deliverable, such as a study guide, comparison table, CSV, timeline, or one-page brief.

Premium and Laureate for Teams

Teams is now two seat plans: Premium for Teams and Laureate for Teams.
Every seat gets that plan's models, features, and limits, plus a shared library, roles, and one bill.
A stronger personal plan still takes priority, and there is no team credit wallet.
Start with 1 to 29 seats, billed monthly or yearly.
Admins get a Team page with Overview, People, Library, and Plan, plus a 7-day activity view.

Simpler AI model controls

Chat, Research, and AI creation now share one consistent model selector.
Auto, Auto (Premium), and Auto (Laureate) are visible on every plan.
The Frontier Boost toggle switches Auto (Laureate) between Claude Sonnet 5 and GPT 5.6 Luna.
Laureate's 75 AI Creation Credits, 10 creation Frontier Boosts, and 20 Chat Frontier Boosts now reset weekly instead of monthly.

Smaller changes

New Deep Research setups start at Medium depth, and a running research plan is now easier to tell apart visually.
Creation setup panels now use consistent, larger section titles and descriptions.
The upgrade screen now lists Export & download and Priority support among the Premium benefits.
The flashcard Sources tab is now AI & Sources: create a podcast, video lecture, study guide, and more from the deck's sources.
"Created from this" lists now update as soon as a generation finishes, with no page reload needed.
Premium in India is now ₹1,299 per month or ₹6,199 per year, saving about 60% on the yearly option.
YouTube videos with captions are no longer rejected as "no captions available" when the transcript service is briefly busy.
Long YouTube videos without captions now finish transcript processing reliably.
Anki import now supports decks exported by Anki 2.1.50 and newer, and corrupt decks get a clear explanation.
Videos without an audio track now import normally instead of failing, with transcript-based features unavailable.
Chat replies that hit the four-image generation cap now finish with the images already created instead of failing.
Cmd+K and Ctrl+K open workspace search ready for typing, and results show content and chats together without layout shifts.
Flashcard decks now show how many Questions are ready directly in the tab.
Signing out or switching accounts in the same tab now fully clears the previous account's library and session data.
After sign-up, a loading indicator now stays on screen instead of a blank page flashing before the welcome flow.

August 10, 2026

Dedicated podcast, video, and flashcard pages in Explore

Explore now has public pages for podcasts, videos, and flashcards, featuring real community content in English and Japanese.
Podcasts get a featured player and a ranked episode chart.
Videos get a cinematic billboard and a thumbnail wall.
Flashcards get an interactive sample card you can flip.
Podcast and video items across Explore now show episode duration and chapter counts.

Reference earlier chats and your full library with @

Type @ in Chat to find a previous conversation and bring its user and assistant messages into the current discussion.
Typing @ in Chat now searches your complete library, including items that have not been loaded yet.
Older pages, files, recordings, podcasts, Research sessions, and folders are now easy to find with @.

Smaller changes

Upgrade and download dialogs now close when you sign out or switch accounts, so the next account never inherits the previous session.
AI Meeting Notes mentions in Chat now use their red recording accent instead of appearing as a generic gray upload.
AI Video Lectures keep signed values from sources, leave production notes out of narration, and reliably include requested slide decks.
Returning to a Chat now restores a completed answer.
Document paging now reaches the final page and has accessible controls.
AI creator instruction counters and upload limits now explain exactly what is blocking the next step.
Free Exam screens now explain the one-attempt daily allowance and when it resets.
PDFs chosen from a folder's Import screen now go through the normal uploader and save back into that folder.
Images uploaded to a folder now remain visible there.
File pickers now match the formats the uploader supports, and cloud-synced files show a clear retry message if access is lost.

August 9, 2026

Companion slide decks for every video lecture

Companion slide decks are now included for Free members too.
Decks are enabled by default for every Standard AI Video Lecture.
Everyone can review and present the deck in the new Slides tab on the video lecture page.
Downloading decks remains a Premium feature.

Smaller changes

Fixed Word and PowerPoint files over 32MB failing to convert in chat and Google Drive imports. Files up to your plan's limit now work.
Opening Home directly now loads your recent chats and workspace instead of sometimes showing the new-user empty screen.
Removed the redundant AI Generated badge from video lecture pages.

August 8, 2026

Video lectures get new palettes, Dark Mode, and a deck

Standard video lectures draw from a pool of hand-tuned color palettes, so each new lecture gets its own look.
Charts, cards, and embedded diagrams follow the palette drawn for each lecture.
A new Dark Mode toggle in the video creation panel renders every slide on one of five dark palettes.
Every Standard AI Video Lecture can include a companion PDF deck, enabled by default on every plan.
The deck opens in its own Slides tab without using another AI creation; Premium members can download it beside the MP4.

Calmer podcast and video lecture pages

AI Creations and source lineage now share one AI & Sources tab across podcasts, video lectures, PDFs, files, and Research.
Podcast Producer Notes and Related Resources are combined into one Deep dive tab, and video Questions moves into the tab row.
Chapters are now a numbered list you can scan and jump from, replacing the sideways-scrolling cards.
The episode header is one details line with a two-line summary, so the slide, tabs, and chapters fit without scrolling.
Podcasts with chapter slides gain a synchronized Video mode and a share-ready MP4, with MP4 downloads on paid plans.

Stronger reasoning in paid Auto

Paid Auto now uses high reasoning effort in Chat and agent-powered AI creators.
This covers Research, slides, documents, video lectures, infographics, and podcasts.
Flashcards keep their balanced medium setting, and Free Auto is unchanged.

Smaller changes

The bottom player now stays pinned while signed-out listeners browse a shared podcast episode.
Fixed chat turns that could end with a “used all of its research steps” error; the assistant now answers from gathered results.
Stopping a new Chat answer now works reliably even while the stream is still starting.
Switching away from the Library tab in an AI creator now keeps the folder and page you were browsing.
AVIF images imported from a URL are now recognized correctly.
The docked podcast player keeps its chapter label aligned after a chapter seek.
Study progress is now cleared and isolated when accounts change.
Older study-guide paragraphs render again instead of disappearing.
Pill-style selectors now keep their fully rounded shape at every size.
Shared flashcard study guides now check that the deck belongs to the page being viewed, keeping other decks' highlights private.
Settings, sharing, and other dialogs now open above the entire page, including an expanded podcast chat.
Downloads of Chat podcasts, file and Office preview fallbacks, and Image Studio images now consistently require Premium.
Free members now see the share-or-upgrade choice instead of downloading directly.

August 7, 2026

Weekly AI Creation Credits on Premium

Premium includes 10 AI Creation Credits each week, shared across flashcards, podcasts, Deep Research, and the other AI creators.
The weekly allowance resets every Monday.
Settings → Billing now shows your live weekly usage.
A failed creation returns its slot automatically.
At the limit, all existing work stays available and the next reset is shown clearly.

Podcasts get Producer Notes and a docked player

New podcasts include timestamped Producer Notes that add researched context at the moments where it matters.
Jump to a note's exact point, follow its cited links, or send it straight into the podcast chat.
A Related Resources section curates trustworthy websites, papers, archives, courses, and exhibits to explore after the episode.
Listening check-in questions are now created alongside the other extras while the audio records.
The player docks in a bar at the bottom of the page, keeping play, skip, speed, and your position in reach.

Home opens with your next session

A continue card at the top of Home names your next step: cards due with a time estimate, a deck in progress, or your latest material.
After a few days away, Home greets you with what's waiting instead of a welcome-back popup.
Create with AI shows the most-used tools first and moves the rest under More.
"Jump back in" is removed, and Folders now appear below your work once you have one.

Smaller changes

Creation tiles and chips on Home and in the upload flow now share one flat button style in light and dark mode.
Folder pages now show just the name, search, New, and your content; search filters items and subfolders as you type.
Every AI creator now offers format-specific expert styles, and quiz styles and custom instructions now steer generated questions.
Chapters, Summary, Transcript, Tasks, AI Creations, and Sources now share one card layout across content pages.
AVIF images can now be uploaded anywhere images are accepted, including flashcards, chat, and pages.
Free Study now reliably stops after five questions, and its Questions count shows the real daily allowance.
Flashcard controls now stay single and readable before and after a card is flipped.
Pages now open directly as one editor without an extra AI Creations tab.
Pasting from Google Docs, Word, and other rich editors into pages keeps headings, formatting, links, lists, quotes, and tables.
Two-host podcasts now stay at exactly two voices, and delivery directions no longer leak into transcripts or speech.
Finished podcast and video audio is now mastered to a consistent listening level.
Fixed long account emails in Settings, narrow flashcard tabs, and a false content-loading error on signed-out Library visits.
Library search in every AI creator now matches Cmd+K, showing newly added sources and loading results beyond the first page.
Video Lectures keep narration at its natural pace in Freestyle, Illustrated, Math, Kids, and Short modes.
Video lectures that hit their AI time limit near the end are now reported accurately and retried when possible.
Importing a file by URL now works with slightly misconfigured site certificates and explains clearly when one can't be trusted.
PowerPoint previews now show a clean, scrollable deck with every slide fitting the panel in light and dark mode.

August 6, 2026

More reliable study guide creation

Study guides and other AI documents now retry automatically after a temporary startup problem instead of ending the creation.
Word and PowerPoint study guides now consistently name the original document.
Native spreadsheet files no longer lead to a PDF-only summary page that cannot open them.

Smaller changes

Flashcard read-aloud now starts reliably in Safari instead of downloading the audio into a silent player.
Changing a page heading into a paragraph now updates the block immediately and stays correct after reloading.
Selected text and selected blocks on a page can now be deleted normally.
The profile menu no longer repeats team setup or deleted-item shortcuts; deleted content is still under Library → Deleted.
Profile pictures now stay circular when the sidebar narrows.

August 5, 2026

Chat replies that stay in place

Typed, retried, voice, and Help chat replies now update in place from the first word to the finished answer.
Formatting no longer shifts when a reply finishes streaming.
Chat replies keep generating when you visit another page and are still live when you return.
You no longer need to refresh or restart to see a reply that was generating while you were away.

Video Lectures and short video quality fixes

Video Lectures now repair or reject narration that stops mid-sentence before recording begins.
Completed Video Lectures now finish every explanation cleanly.
Short videos restore their original Scholarly art direction and word-for-word paper-strip captions.
Short videos use four distinct, completely still illustrations.

Smaller changes

Fixed daily Chat allowances not resetting consistently, which could block a new day's messages while Billing showed unused credits.
Shared flashcard decks now show that autoplay requires sign-in instead of offering a control that cannot play while signed out.

August 4, 2026

Deepseek Flash in every model picker

Deepseek Flash is now available as a new model option.
Available in Chat, Research, Slides and documents, Video Lectures, Infographics and Mind Maps, Flashcards, and Podcasts.
It is an efficient, low-cost reasoning option.
It is text-only, so visual source workflows use the text extracted from your material.

Simpler model pickers

Only Auto (Premium) and Auto (Free) are shown up front.
Manual choices now live under Premium Models and Free Models.
Lock and speed icons no longer clutter the list.

Stop & send for queued chat messages

Queued chat messages now have a Stop & send option.
It interrupts an outdated response and safely settles any partial answer.
Your next message is sent right away without another click.

Smaller changes

Next and Back in creation modals now switch steps immediately, without a blank pause.
Fixed refining an existing flashcard deck failing before the request could be sent.
Chat streaming, read-aloud, and flashcard autoplay now keep going if your session briefly expires.
Fixed math formulas creating nested scroll areas in Chat and other generated content.
Podcasts now treat instructions like “two clearly labeled speakers” as a strict two-voice format, so no producer slips into the audio.
Added sources are now named as a Word document, PowerPoint, spreadsheet, or PDF to match what you selected, not always as a PDF.
Welcome and pricing copy now speaks to learners, researchers, professionals, and teams, not only students heading to class.
Teams pricing now consistently leads with $45/seat/month and lists $324/seat/year as the separate discounted annual option.

August 3, 2026

A voice picker you can hear before you choose

Choosing a voice now opens a full picker instead of a cramped dropdown.
Browse and search all 26 voices, and play a short demo of any voice before you commit.
The same picker is used in chat settings, voice mode, read-aloud, podcast guest voices, and video narration.

Long podcasts that fill the time you ask for

45 or 60-minute podcasts now fill the requested time instead of trailing off around the half-hour mark.
Long episodes are planned chapter by chapter and written in sections.
Episodes that come out far over a shorter requested length are now tightened before the audio is recorded.

Separate Chat and AI Creations settings

The Scholarly AI settings tab is now split into two tabs: Chat and AI Creations.
Chat holds suggestions, auto-play audio, your voice, language, custom instructions, and memories.
AI Creations holds custom designs, check-in questions, and autocomplete.
Video and podcast check-in questions can now be turned on or off from settings, not only from the player.

Smaller changes

The AI Creations and Sources tabs now match the folder studio, with quieter create buttons and simpler one-line empty states.
Tidied the podcast page: one title instead of two, a cleaner details line, shorter tab names, and calmer chapter visuals.
Fixed chapter arrows on podcasts and video lectures flipping to the neighboring chapter and snapping back.
AI Slides no longer invents a date, weekday, event, or causal claim that the source says is not provided.
Chat now finds PowerPoint content, including speaker notes, on the first question instead of saying the deck is unavailable.
Podcast transcripts now show the speaker name at every chapter opening, even if the same person closed the previous chapter.
When an AI Creation Credit limit blocks the next step, Next is now visibly disabled and the upgrade action stays available.

August 2, 2026

Projects

Projects keep the sources, chats, and AI creations for one goal together, with project instructions that every chat follows.
Chats in a project are grounded in its material, and anything you create there is filed back into the project automatically.
Turn any folder into a project in one click, move chats and content in or out from row menus, or drop files onto a project.
Share with your team, by email invite, or by link: recipients see sources and AI creations read-only, and your chats stay private.
Project memory shows what Scholarly has picked up in a project, recalled only there, and lets you retire anything out of date.

A simpler sidebar and creation menus

The sidebar now focuses on New, Home, Chat, and a new Library link, with one Recent list mixing your latest content and chats.
Recently Deleted, Changelog, and Share Feedback moved into your profile menu.
Timeline, SOP, flowchart, lesson plan, and outline now live in AI PDF Creator; pick the document type after adding sources.
Spreadsheet and the former AI File Creator are now one Document & File Creator, while AI Essay Writer stays separate.

Smaller changes

Find projects from Search or Cmd+K, and browse projects and their folders when choosing sources for a new creation.
Expert style cards in every AI creation panel now sit near the top with the other creative choices, instead of at the bottom.
"Number of flashcards" moved from the preset box into Settings, next to language and model.
Free Video Lectures now run up to 2 minutes and free Podcasts up to 5 minutes, with limits and an Upgrade link shown when customizing.
PowerPoint and Word uploads now show their own name and file type everywhere, including viewers and the Sources list, instead of PDF.
Fixed podcasts, video plans, and the help bot saying a PowerPoint's speaker notes were unavailable when they had been extracted.
The "How was this…?" feedback card now only appears on the page it is asking about.
Timeline plan-limit errors now include a direct upgrade action, and opening plan options clears older error banners.
Fixed a temporary autocomplete sync failure reducing your remaining allowance until you reloaded the page.
People viewing a publicly shared page can now export or preview it; private pages stay blocked from public download links.
The Premium offer shown after leaving checkout is clearer, and its subscribe and dismiss actions stay visible on phones.
Podcast errors for unreadable sources now name the problem with each source and what to do, instead of "no usable content".
Expired or revoked Google Drive access now prompts you to reconnect Google Workspace in Settings instead of showing a technical error.
Excel workbooks with native charts now open instead of showing a preview error, with every populated worksheet available.
AI-created spreadsheets now avoid cramped wrapped labels and repetitive chart labels, so dashboard sheets are easier to read.
Podcasts that request a two-speaker format now stick to two speakers, so a producer voice no longer slips in.
New students now see the tools and choose what to study before any pricing prompt.
Upgrade actions now consistently say Premium, and voice limits clearly distinguish Free from Premium.
Free Study shows its 10-card question allowance without interrupting the first question, and explains the upgrade path at the limit.
Workspace pages on desktop once again show the Scholarly logo and breadcrumb separator.

August 1, 2026

Plan any AI creation with chat

Every AI creator now has a Plan with chat button beside its standard next step.
The planning composer opens in place so you can discuss your sources and shape the result before creating.
Planning works for flashcards, quizzes, slides, videos, podcasts, documents, study materials, and more.

Complete flashcard deck duplication

Duplicating a page with a flashcard deck no longer drops cards when an image fails to copy.
A card whose image can't be copied now keeps its question and answer text.
Duplicated decks keep the original card order instead of moving cards with images to the end.
Copied cards keep their AI hint and difficulty label.
Source citations on copied cards open the duplicated PDF or video in your new page, not the original.

More dependable AI creations

AI creations retry or switch to a backup model when a provider returns a malformed response, instead of failing.
A creator no longer stays on "Creating your…" forever; it tells you what happened and hands the page back.
Flashcard decks and quizzes finished in the background stay on Home with a link straight into them.
New decks appear under Jump back in right away, with no page refresh needed.

Smaller changes

Fixed the flashcard feedback prompt covering Next after card five, so longer decks advance normally.
Finished short Video Lectures no longer open a sharing dialog over the Summary, Transcript, AI Creations, and Sources tabs.
Pressing Next in Add content after reaching the free upload limit now opens your plan options instead of doing nothing.
Chat now treats PowerPoint and Word uploads as their original file type, not PDFs, and can find a presentation's speaker notes.
Connecting Google Drive no longer fails if you spend more than ten minutes on Google's account and permission screens.
Shared chat links with a malformed address now show a clear not-found page instead of a generic error.

July 31, 2026

One AI PDF creator for any document

The separate SOP, flowchart, lesson plan, and outline creators are now a single AI PDF creator.
Add your sources once, then pick the document type in the customize step.
Pick "Anything" and describe the document in your own instructions, like a briefing, checklist, or meeting minutes.
Four new document types: cheat sheet, report, case study, and rubric, each source-grounded and typeset like the existing kinds.

New AI File Creator

Turn your sources into a downloadable file: an Excel workbook, CSV table, JSON export, Markdown document, or calendar file.
Describe what you want and it picks the matching format, or start from one of the format tiles directly.
The existing AI Spreadsheet creator is unchanged and is also available as a tile inside the new creator.

Preview Office files directly

Word documents, Excel workbooks, and PowerPoint decks now open in the file preview from chat, research, and saved files.
New Word and PowerPoint uploads keep their original file for preview and download.
Office files over 25 MB show a preview-limit message and a direct download action instead of loading indefinitely.
Fixed spreadsheets containing an Excel table, including AI File Creator workbooks, failing to preview.
PowerPoint decks follow light and dark mode, and trackpad scrolling no longer runs away or triggers back-navigation.

Smaller changes

The upgrade modal now leads with the monthly equivalent for yearly plans (for example $12/mo) and shows the annual total below it.
Free practice exams now use at most one question per prepared flashcard and show that maximum at every exam entry point.
Asking to preserve every source question now requires every source section to finish before the deck is delivered.
Long video-lecture scenes now introduce teaching points throughout the narration instead of showing a static slide early.
Short Video Lectures no longer read source markers or machine-style identifiers aloud in narration.
Research grounded in a linked PDF now draws excerpts from across the document, not just its opening section.
Source-grounded chat tables now cite every factual cell, so each number, date, or quote can be checked in one click.
PowerPoint speaker notes now stay with their matching slide in chat and AI creations.
Infographics now honor landscape and portrait requests, treat historical year ranges as timelines, and flag oversized requests early.
Podcast expansion now catches near-duplicate chapters while keeping distinct chapters that share a generic title.
Fixed uploads failing with a raw "Retry limit exceeded" message. Brief storage hiccups are now retried automatically.
Chat now says plainly when a web search comes back empty, and failed turns no longer save a placeholder reply or spend a chat credit.
Chat reasoning and tool actions now stay in order while streaming, and charts appear as soon as the assistant mentions them.
The Ask, Copy, and Speak menu now stays beside selected chat text, even after scrolling a long conversation.
Undo and redo on pages are now saved, and a related editor crash that stopped further typing from saving is fixed.
Read-only and shared pages now stay read-only, and text typed while blocks are selected is no longer discarded.
Signup and other simple pages load faster because they no longer load the page editor.
Fixed switching from monthly to yearly billing so a simultaneous billing update can no longer roll back your renewal date.
Abandoned-checkout recovery now describes what happened accurately and keeps yearly Premium as the primary option.
Fixed rare errors where a page load or request failed once with a server error and then worked on retry.
Flashcard creation now honors the selected difficulty, and its footer wraps cleanly when a plan-limit message appears.
Creation flows now wait for image uploads to finish, and temporary source-loading failures are retried instead of marked unsupported.
Laureate members now see the correct weekly flagship-credit reset guidance and can receive subscriber feedback prompts.
Research preparation and podcasts now handle missing content safely, and screen readers announce the welcome-back dialog title.
On mobile, interactive podcast questions scroll into view when they appear and stay open while you tap around.
Fixed flashcard study card controls and long questions being cut off on smaller phones.
On mobile, the creation customization step shows a loading state, and Google Drive picks scroll when selecting several files.
"Other Models" now expands inside the model picker on phones, so every model is reachable.
Plan details that appeared only on hover, like Laureate's weekly credit allowance, are now readable on touch.
On mobile, tapping an image to enlarge it now fills the screen instead of opening at half width.
Bigger tap targets on mobile across study, exams, saved instructions, sharing, and the plan billing toggle.
Blog posts, file page tabs, search filters, and marketing stats now lay out properly on phones.

July 30, 2026

AI Creations tab for every source

Pages, PDFs, videos, AI Meeting Notes, podcasts, research reports, and saved files now have an AI Creations tab.
New flashcards, quizzes, podcasts, slides, study guides, essays, and more stay linked to their source, including the PDF pages used.
Search and filter a source's creations by study tools, media, or documents, then open any result directly.
Flashcard creation for PDFs, videos, and AI Meeting Notes now lives in AI Creations, replacing their standalone Flashcards tabs.
New decks, documents, files, video lectures, and podcasts have a Sources tab that links back to the original material and PDF pages.

Auto chat now uses GPT-5.6 Luna

Auto now uses GPT-5.6 Luna for chat on every plan.
Premium and higher plans use high reasoning effort for stronger answers.
Free chat uses Luna with its efficient reasoning setting.
AI creators keep their existing automatic models.

Smaller changes

Creation shortcuts moved from the assistant's empty conversation to AI Creations, keeping chat focused on questions and explanations.
Fixed videos opened in a background tab showing an unnecessary Retry prompt before the tab became active.

July 29, 2026

Grok Voice Think Fast 2 for the voice assistant

Scholarly's voice assistant now uses xAI's Grok Voice Think Fast 2.
Voice conversations respond faster.
The assistant follows complex study requests more reliably.
Interruptions and tool use are handled more naturally during conversations.

Smaller changes

Returning to Home from PDFs, videos, flashcards, podcasts, recordings, research, and shared chats is now faster.

July 28, 2026

Podcasts with chapter slides

Every new podcast now includes a visual companion by default, for everyone.
Scholarly generates one Editorial slide per chapter alongside the audio.
Slides follow the current chapter automatically as you listen.
You can turn slides off for an audio-only episode.
Previous and Next now switch chapter slides immediately in Safari, even while the audio is paused.

Choose the AI that writes your podcast

Podcast creation now includes an AI model picker.
Free uses Gemini 2.5 Flash; paid plans default to Gemini 3 Flash Preview and can choose either.
The selected model writes and reviews the transcript only; voices and audio generation are unchanged.
Every podcast speaker now uses the exact name of the Grok voice you hear, such as Cosmo, Ursa, or Sirius.

Better recovery from brief connection problems

Video and audio downloads now retry after a brief connection drop instead of failing immediately.
Interrupted PDF page image uploads are now retried before processing starts over from scratch.
Accounts left open in an older browser tab no longer crash while loading after Scholarly updates available AI models.

Smaller changes

When AI Chat creates content or research, it now uses Auto (Free) on Free and Auto (Premium) on paid plans, matching the creation dialog.
Named AI models now use compact single-line rows in every picker; Auto (Premium) and Auto (Free) keep their explanations.
Podcast interactive questions now have their own counted tab, and longer podcast pages scroll all the way to every panel.
Feedback cards no longer cover quiz or exam answers, stay out of focused study routes, and can always be dismissed.
Flashcard quantity options now describe AI-generated counts as targets rather than guarantees.
AI Slides now checks card spacing and complete equations before finishing, reducing overlapping text and missing formulas.
Text fields no longer grab keyboard focus automatically when pages, searches, or study steps open.
The welcome tour now splits its introduction, plan comparison, study tools, and start choices into four focused screens.
Completed flashcard tasks now offer Start studying directly from the completion message and Home task tray.

July 27, 2026

Broader, more honest practice exams

Exam mode now shows how many of your deck's cards have practice questions, as "N of M cards prepared".
On paid plans, creating an exam fills in only the missing practice variants needed for that exam's size.
Paid exams are no longer restricted to a stale 10-card pool.
Free decks show the 10-card preparation limit where you pick an exam, with the upgrade path next to it.

Video lectures grounded in your source again

Prepared video lectures can once again use PDF pages with figures, tables, or equations, so those details make it into the lecture.
Video lectures can now show concise on-screen cards instead of repeating the narration.
Scenes without cards still use the existing narration-based text.

AI creations start in your saved language

Every AI creation now starts in your saved language.
The first time, Scholarly uses your browser locale, with English as the fallback.
Changing Language in any creator updates your account preference everywhere.

Smaller changes

Google Drive files that can't be downloaded are now flagged when you pick them, instead of failing partway through a generation.
Drive files whose owner turned off downloading now explain that and how to fix it, instead of asking you to reconnect your account.
PDFs now open reliably in the in-app reader instead of failing while the viewer loads.
Documents with no lesson content now point you to the pages or link that have the material, instead of suggesting you re-save as PDF.
Opening plan options from an Upgrade Plan button no longer shows a limit warning just because free credits were used.
Removed the bright outline around the upgrade dialog when keyboard focus moves into it; buttons and fields keep their focus indicators.
AI Slides now keep source values and units exactly, leaving a column out of dense tables instead of inventing summaries or units.
The welcome tour can now be skipped from its first screen and dismissed with the Escape key.
The cancel-flow pause offer now allows up to 3 months, matching billing settings (it was previously capped at 2).
If you canceled to avoid auto-renewal, the follow-up email now offers a plain restart instead of a discounted subscription.
Win-back discount codes now stay valid for 21 days after paid access ends, instead of sometimes expiring before access ended.

July 26, 2026

Flagship credits for Scholarly Laureate

Laureate includes 2,000 weekly credits for flagship-model work.
Standard and advanced models stay unlimited on both Premium and Laureate, so Premium access is unchanged.
Laureate members can see their flagship-credit balance, usage details, and reset time in Settings → Usage.
Laureate members can top up flagship credits in a heavy week without changing plans.

Scholarly for Teams, easier to find and start

Teams is now a full plan on the pricing page, with per-seat pricing that follows the monthly/yearly toggle.
The profile menu, billing settings, and plan options now open team setup directly instead of a marketing page.
Signing up or using Scholarly with a work or school email now shows Teams on signup and your home page.
"Talk to sales" on the Teams page is now an on-page form, so you no longer need a mail client.
The Teams page now leads with turning handbooks, recordings, and decks into onboarding and training content.

Fewer failed generations

You can now make a video lecture from a short topic like “Krebs cycle” instead of being turned away.
Content-safety refusals now retry with another provider, so medical, anatomy, and clinical sources no longer dead-end.
A PDF with nothing readable now says so right away instead of showing “still processing” for ten minutes.
Google Drive files over your plan's size limit are flagged as soon as you pick them, with the upgrade path shown.
Video lectures whose picture ends before the narration are now re-stitched during generation instead of failing at the end.

Smaller changes

Scholarly's paid plan is now called Premium everywhere in the product; your subscription and billing are unchanged.
PDFs and videos now show the flashcard deck made from them, and "Jump back in" on the home page includes flashcard decks.
The free-limits banner now waits until your first creation is ready instead of appearing while it generates.
The chat limit now tells you when it resets.
Finishing a creation no longer flashes the create form with a "Retry" button while your new content opens.
PDF and video summaries now follow the language of your source by default instead of forcing English.
LaTeX in flashcards and study questions no longer loses commands such as fractions, Greek letters, or text labels.
AI creations are now told which PDF pages were skipped so they can read them instead of silently missing part of the source.
Flashcards now stop when only a link or reference to missing material arrives, instead of inventing a deck.
AI podcasts now keep a direct, appropriate tone for sensitive human topics instead of forcing cosmic metaphors.
Infographics no longer risk printing internal layout labels such as “TITLE” or “SECTION” on the finished poster.
Fixed switching a monthly subscription to yearly billing, which always failed with a "please try again" message.
Fixed some new accounts skipping onboarding after signup and being marked as onboarded without seeing it.
Brief server errors no longer send you back to the login form or sign you out while your session is still valid.
A failed page navigation no longer reloads the app and discards an edit being saved; reloads only happen for new app versions.
Fixed scholarly.so/content showing a "Page not found" error instead of your library.
Reloading a podcast part-way through no longer re-asks questions you already answered.
Video generation now rebuilds damaged files, no longer stalls for about 20 minutes, and no longer restarts nearly-finished lectures.
Explore search now retries when results are briefly unavailable instead of falling back to a weaker match.
Requests in progress during a brief outage are now allowed to finish instead of being cut off.
Dialogs now move keyboard focus inside when opened and return it when closed, for keyboard and screen-reader users.
Pressing Enter in the signup email field now moves to the password field.
Video lecture title and closing cards no longer show an internal authoring note or cut off lines that still fit.
Fixed the hard-to-read “NEW” badge in light mode and compact library rows blending into the page.
Help and marketing pages no longer list slide themes that are no longer offered.
The create screen no longer says “Your content is ready!” before your file has uploaded.
The monthly and yearly options in the upgrade dialog can now be switched with the arrow keys.
Podcast instructions that name specific hosts are now followed instead of always using the standard three-host cast.
Audio and video playback errors now say what actually happened instead of blaming an expired link.
Help lists and navigation no longer pause unexpectedly or reload the page you were reading.

July 25, 2026

Interactive questions in podcasts

New podcasts now include quick check-for-understanding questions while you listen.
Each question appears right after the idea it tests finishes, and never pauses the audio.
Answer to get an explanation grounded in the episode.
Revisit any question from the episode's question list and track your score across attempts.
You can switch questions off from the player at any time.

Real figures in AI Video Lecture scenes

Chart, equation and statistic scenes now use the real figures from your source.
These scenes no longer fall back to a generic two-bar chart or the scene title typeset as an equation.
Scenes with no figure worth showing now render as a clean text callout instead of an empty or invented visual.

Exams open on any deck

The Exam tab now prepares its own study questions instead of waiting on the Questions tab.
Exams now open on decks you have not quizzed yet.
If questions cannot be prepared, you see a clear message and a Try again button instead of an endless spinner.
After using your daily practice questions, a deck now shows the full limit screen with questions remaining and reset time.
The limit screen also offers a way back to your flashcards.

Smaller changes

AI Chat and creation dialogs now offer Auto (Premium) and Auto (Free) to use the recommended model for your chosen quality tier.
Voice-picker search now uses one consistent surface in both light and dark mode.
Plan options now show the yellow limit explanation only when the action that opened them actually reached that limit.
Locked Full HD and custom-length video options now explain what Premium unlocks before showing plan choices.
Welcome tours now introduce what Scholarly can do before offering to skip the rest.
The upgrade dialog now scrolls your selected plan into view on short screens, so the highlighted term matches the button price.
Fixed the Home chat box on phones overlapping its own buttons, so tapping anywhere in the field starts typing.
The “How was this?” card after a generation no longer covers the chat send button.
Copying a share link from a private item now says up front that it will make the item viewable.
Home creation shortcuts are back to the three-card layout with full AI tool names and the original quick-access order.
If Google's permission screen closes without the Google Drive box ticked, Scholarly now says which box to tick and offers Try again.

July 24, 2026

Teams for training, onboarding, and sales

New Teams pages cover employee training, employee onboarding, sales enablement, and businesses.
Teams seats cost $45/seat/month or $324/seat/year.
New help docs cover getting started, billing, sharing with Your team, and switching from an individual plan.
Teams help now explains Premium usage and limits and the shared library workflow for L&D and research groups.
A new ChatGPT Business (formerly Team) comparison and buyer guides cover AI employee training tools and sales enablement.

Video lectures and research recover instead of failing

Interrupted video lectures and research now retry automatically instead of failing and refunding.
Videos that fail a final quality check regenerate once, and corrupted or silent renders are caught while they can be rebuilt.
Long video lectures no longer stop early, and lectures stuck retrying one step now get corrected or retried automatically.
PDF study guides no longer fail or get cut short on longer documents.
Flashcards from dense sources arrive far sooner, and study materials no longer fail when the model writes slightly more text.

A much better phone experience

Main buttons stay reachable on the post-upload chooser, post-purchase welcome steps, checkout offer, and practice exam setup.
Long flashcard answers, quiz questions, and session summaries now scroll, and long multiple-choice options show in full.
Plan and competitor comparison tables stack into readable cards, with previously hover-only explanations shown inline.
Video, recording, and PDF pages now show the item's name, including when you open a shared link.
Reaching your AI Creation Credit limit shows upgrade options, and the yearly checkout reminder no longer covers chat or recording.

Smaller changes

Updated PDF Summarizer, flashcard, homework-helper, Quizlet, and Chegg pages with clearer comparisons, examples, and accessible imagery.
Attachments sent with your first chat message now stay on that message instead of moving to the next one.
Multiple images in chat now show in a compact equal-height grid (up to four) instead of a tall list of large previews.
Fixed fresh web searches in AI Chat failing, so they return current sources again.
Chat replies no longer get stuck when a model rejects a long conversation; the message moves to another model instead.
The chat model picker now highlights four recommended choices.
A temporary connection or server error while loading your account no longer signs you out.
Recording and transcription no longer use an AI Creation Credit.
When one source fails while creating Flashcards or a Quiz, your other sources stay ready and you can remove the failed one.
Creation dialogs now keep limit guidance and an Upgrade action visible when a plan limit is reached.
Post-purchase tools and cancellation guidance now keep the record of features you have tried after your account refreshes.
Signing up from a tool page now carries that tool through onboarding and opens the matching creation flow.
The welcome tour now introduces upgrading earlier.
Slide creation no longer gets stuck repairing harmless text overlap.
Very High flashcard sets once again target at least 200 useful cards when the source supports it.
If Essay Writer generation is interrupted after a refund is claimed, your AI Creation Credit is now returned.
Google Drive searches for large or shared study materials now read every page of results.
Your active plan now appears immediately after a payment or plan change instead of briefly showing the old plan.
The plan chooser can now start with monthly billing selected, with yearly savings still visible.
Monthly members see the yearly-switch reminder near renewal, at most once per billing cycle.
Leaving a yearly checkout now shows a recovery screen that leads with monthly billing.
Dialogs no longer pre-select Close, Skip, or other buttons when they open, removing the unexpected blue outline.
Creation settings now sit in their own labeled section, separate from the design choice, with one option per row.

July 23, 2026

Reliability fixes for chat, PDFs, video, and teams

Long, tool-heavy AI Chat conversations no longer occasionally fail before the answer starts, and your latest message stays intact.
Large PDF summaries now finish reliably instead of occasionally failing near the end.
Healthy long video lectures are no longer rejected over a harmless sub-second gap between audio and video timestamps.
Team workspaces load more reliably when many people sign in at once, with fewer account-loading failures.

Smaller changes

Refund policy clarified: most requests within 30 days are approved; unusually extensive use, fraud, or abuse may need case-by-case review.

July 22, 2026

Faster processing for large textbooks and PDFs

Large textbooks and PDFs now finish processing much faster, so you can start studying sooner.
Large PDFs now open incrementally instead of downloading and rendering the entire document up front.
Large PDFs stay responsive while you scroll or zoom.

Longer custom instructions and Claude Sonnet 5

Custom instructions now allow up to 5,000 characters.
The 5,000-character limit applies across Chat, Skills, Custom Designs, and every supported create flow.
Plus and Premium members can now choose Claude Sonnet 5 for AI Video Lectures.

Smaller changes

AI Study Guide is easier to find on Home and after uploads, and Worksheet now lives under More in a balanced action grid.
Fixed generated and edited page titles reverting to Untitled when you leave the editor.
Plan dialogs now show the true reason they opened and support complete Tab and Shift+Tab keyboard navigation.

July 21, 2026

New Gemini models

Gemini 3.6 Flash and Gemini 3.5 Flash Lite are now available.
Both models work in Chat, Research, Slides, Video Lectures, Infographics, Mind Maps, and Flashcards.
If a retired model was selected, Scholarly now switches to the current default for that feature instead of showing a different model name.

Clearer plans, limits, and checkout recovery

Yearly and monthly checkout options once again show the exact charge up front, such as $144 today or $30/month.
You can choose a monthly plan directly at checkout.
After checkout, Scholarly shows the file, action, or limit that led you there so you can pick up where you left off.
Upload and AI limits stay visible on Home, Chat, and page upload surfaces with accurate reset or lifetime wording and an upgrade action.
Uploaded files are now saved before customization or checkout, so returning to the library no longer loses them.

Smaller changes

Practice Exams in Auto language now follow the main language of the source instead of a small translated glossary.

July 20, 2026

Scholarly for Teams now starts at one seat

You can create a full Teams workspace for just one person.
Add teammates later as your team grows.
The 29-seat self-serve maximum is unchanged.

A refreshed Help Center

A new Memory guide explains how the AI Assistant saves useful facts.
The guide also covers adding memories yourself and marking them outdated or restoring them.
Every article was checked against the current product, with outdated model, plan, workflow, and feature details corrected.

Chat keeps going when your connection drops

Chat keeps generating when a phone locks or a connection drops.
Chat reconnects quietly once you are back.
Stop is now an explicit action instead of an accidental disconnect.

Smaller changes

Fixed emoji in Word, PowerPoint, and other Office files disappearing from converted PDFs.
Fixed clearing the email field on login silently clearing or desynchronizing the password field.
Podcasts, recordings, and Deep Research now show clear Processing or Failed states across Home, libraries, and search.

July 19, 2026

Mind Maps with real, readable text

Mind Maps now render as crisp PDFs with exact labels.
Mind Map text stays searchable, selectable, and easy to read.

More resilient AI generation

Podcasts save completed chapter audio instead of discarding the whole episode.
Podcasts retry through temporary voice-provider outages.
Video Lecture narration plays as one continuous performance whenever possible, without speed and tone resets at each chapter.
Flashcards can read browser-printed PDFs from their page images when the selectable text has only headers, dates, and URLs.

Smaller changes

Email signup or login now offers a Continue with Google button when the address belongs to a Google-linked Scholarly account.
Branded Short videos now include a crisp Scholarly mark, while paid no-logo exports stay watermark-free.

July 18, 2026

Smoother streaming in AI Chat

New answer text fades in smoothly as it streams.
Citations appear as source badges while the answer is still streaming.
Finished responses keep their layout instead of snapping into a different one.

More reliable podcasts

Podcasts retry temporary transcript and voice-generation failures a limited number of times instead of failing the whole episode.
A single provider hiccup no longer ends podcast generation.
Older podcast setups with a missing guest name now use "Guest" instead of crashing during transcript creation.

Smaller changes

Flashcard Export and Daily Goal now open with a click or tap, so PDF, Print, and Anki export work on desktop and touch screens.
Monthly subscribers canceling over price can switch to yearly billing in the cancellation flow, with the full charge shown first.

July 17, 2026

Illustrated design for Video Lectures

Video Lectures have a new Illustrated design: a full-length, hand-drawn explainer on a clean white board.
One quirky character sketches every idea in the video.
Each scene turns one idea into a simple visual metaphor, like a coin ramp, a growing jar, or a balance scale.
Handwritten labels are drawn right into the art of each scene.

Short design for Video Lectures

Video Lectures have a new Short design: a vertical (9:16) video that distills your source into one tight, illustrated story.
Narration appears on screen word-for-word as paper-strip subtitles synced to the voice, so Shorts work with the sound off.
Every Short uses one of four Scholarly styles: cobalt dither, oil painting, papercraft diorama, or sketchbook journal.

Smaller changes

Illustrated and Short videos now hold each scene as a clean still with crisp cuts instead of a slow zoom.
The video player now shapes itself to vertical videos instead of adding black bars.

July 16, 2026

Smaller changes

Flashcard study questions no longer disappear when the AI skips a card; missing ones are filled in next time you open study mode.
PDF summary errors now show a clear, friendly message instead of technical error text.

July 14, 2026

26 multilingual Grok voices across Scholarly

Scholarly now includes 26 official multilingual Grok voices: 21 new voices plus Ara, Eve, Leo, Rex, and Sal.
The voices are available for AI Video Lectures, AI Podcasts, read-aloud, and live Voice Mode.
Choose the exact voice for live Voice Mode and read-aloud in Scholarly AI settings.
Choose a podcast guest voice or a video narrator, and each option explains its tone.
Luna, a gentle and patient voice, is the new default for live Voice Mode and read-aloud.

Smaller changes

For podcasts and videos, leave voice selection on Auto and Scholarly will choose voices that fit your material.
Voice Mode now handles reconnects and interruptions more reliably, without overlapping old audio or losing a turn that needs saving.
Fixed a crash on chat, PDF, and flashcard pages when a streaming answer mixed a plain "*" with a math formula.

July 12, 2026

Smaller changes

Fixed a rare error that could stop the Home page from loading for some workspace members.
Fixed keyboard shortcuts, including ⌘/Ctrl + K search, occasionally erroring on Safari when a password manager or autofill filled a field.

July 11, 2026

Smaller changes

Fixed "Copy share link" failing with "Couldn't copy the link." on Safari and iPhone/iPad.
Flashcard generation now retries on a second model instead of failing the deck with "try different files" when a response can't be read.

July 9, 2026

Refreshed sidebar layout

Home, Chat, and Recently Deleted now sit at the top of the sidebar.
A primary New button and search sit below them.
Content / Chats tabs filter Recent, Folders, and chats.
Recent and Pinned now use the same type-colored rows as Home.
Sidebar text is easier to read in dark mode.

July 8, 2026

Grok 4.5 from xAI

Grok 4.5 from xAI is now available in chat and create tools, replacing Grok 4.3.
On paid plans, Grok 4.5 is the default for AI Video Lectures, AI Slides, Infographics, Flashcards, and Research.
Free plans stay on DeepSeek.
Previous Grok 4.3 selections continue to work automatically.
Grok 4.5 always thinks before answering in AI Chat, so chats no longer fail if thinking was turned off.

Smaller changes

Fixed sign-up sometimes crashing on the last registration step after Google sign-in.
Fixed your chat message disappearing after a failed reply. The original text now stays visible so you can retry.
Refined the yearly Premium savings banner on Home so the discount and switch-to-yearly offer are easier to scan.
Tightened spacing in the Create-with-AI option grid so longer titles like “AI Video Lecture” are no longer cut off.
Creation loading screens no longer show rotating study tips, only the status and progress.
Fixed design and style option cards in create modals flashing a light white hover in dark mode.
The button for closing a create modal while generating now just says Close (no more “continue studying” wording).
The “More credits” model badge now only appears for team and enterprise accounts that use weekly credits, not individual plans.

July 7, 2026

Cleaner sidebar

A prominent New button and search now sit at the top of the sidebar.
Content and Chats sections let you browse your library, folders, and recent chats.

AI video lectures from photos and scans

You can now create AI video lectures from photos and scanned documents.
Scholarly automatically reads your images and image-only PDFs to build the video.
This replaces an unhelpful error that asked you to pick a different model.

Smaller changes

Fixed AI Flowcharts steps showing as solid black boxes that hid their labels; each step now shows its text clearly.

July 6, 2026

More reliable flashcard creation

If the AI returns an empty response, flashcard creation now retries automatically and falls back to a backup model.
A brief AI hiccup no longer fails your deck or uses up a creation.
Fixed a rare issue where flashcard creation could fail with an error or get stuck partway after your cards were generated.
Flashcard creation now reliably saves your finished deck.

Smaller changes

AI video lecture sources with too little text now show a clear "source is too short" message instead of suggesting you run OCR.
Tidied the mobile sign-up and login forms and fixed a rare Google sign-up crash.

July 5, 2026

A cleaner AI Meeting Notes page

The bulky playback panel is now a single compact bar, giving your transcript, summary, and tasks more room.
A refreshed layout adds more breathing room around your transcript, summary, and action items.

Share entire folders

Folders can now be shared just like any other item.
Recipients get a read-only view of everything inside the folder.
Folders shared with “Your team” show up in your team’s shared library automatically.

Smaller changes

Chat now retries automatically when the AI model is momentarily overloaded, instead of showing an error.
The final signup step now always pre-selects the free “Just me” option; the paid Teams option is never pre-picked.
Clearer messages when flashcards can’t be created from a short topic or a photo with no lesson content.

July 4, 2026

Scholarly for Teams

Put a class, department, research lab, school, or company on a single Premium or Laureate plan with shared material.
Every seat inherits the chosen plan's models, features, and limits, and admins control which model tiers the team can use.
Each member uses the limits included with the plan, so there is no shared team balance to divide or watch count down.
Admins can invite by email or link, assign Member or Admin roles, share source libraries, and pay one central bill.

AI reads your PDF pages as you see them

The AI now reads relevant PDF pages as full page images when building slides, study guides, worksheets, videos, and more.
Diagrams, equations, tables, and handwriting are read as they appear instead of relying on extracted text.
Infographics and mind maps now look at your PDF's actual pages when planning, so charts and diagrams make it into the result.

Image Studio upgrades

A new Quality selector lets paid plans generate sharper 2K images on supported models and shows which models will use it.
Each tile now shows which model is producing it while images render or if one fails, so comparing models is easier.
The Style menu is removed; describe the look you want in your prompt instead.
Deleting an image now asks for confirmation first.

Smaller changes

Fixed the Image Studio counter showing “Infinity free left” on unlimited plans.
Light mode polish: panels, cards, and labels across chat, recordings, podcasts, flashcards, settings, and more now have proper contrast.
Filled in missing dark mode styles across the app, and page editor equations no longer render as white strips in dark mode.
Fixed white backgrounds on the billing and connections pages and hard-to-read flashcard statistics chart labels in dark mode.
PDF and slide viewers now have a Print option in the three-dots menu.
Links added as sources no longer use up your free lifetime file upload.
Added a Skip option to the welcome tour.
Quiz and exam questions now focus on the concepts a figure shows instead of exact values read off illustrative graphs.
Fixed “An error occurred while creating the checkout session” blocking some users from upgrading; checkout now opens reliably.
Home and All Content no longer reload more of your library than needed when switching views or filters.
The welcome guide now matches your actual number of free AI creations instead of overstating it.
Fixed pacing markers like “[slow]” occasionally appearing in AI video lecture transcripts.
Uploading a “Print to PDF” of a SharePoint, OneDrive, or Google Docs viewer page now prompts you to download the original file.
The Formula Sheet Generator's Print button now prints just the sheet, falls back to a download if blocked, and works on iPad and iOS.
Several menus and drawers now open instantly instead of feeling sluggish.
Fixed the “last viewed” time on shared content showing an older visit instead of the most recent one.

July 3, 2026

Custom Designs are now free on every plan

Anyone on any plan can now create and use custom designs, which used to be a premium-only feature.
A custom design saves your style instructions plus reference images so you can reuse your own look.
Apply a custom design when generating Slides, Story Books, Infographics, Mindmaps, and AI Videos.

Saved web pages work as study material

Saved web pages added as sources (for example Google Drive or Google Docs viewer pages) now have their readable text pulled out automatically.
These pages no longer fail when you add them as a source.
If a page is only a sign-in or viewer shell with no content, we tell you right away to download the actual file instead.

Smaller changes

Upload Content, flashcard and quiz creation, the welcome tour, and the plans dialog now open instantly the first time.
Opening a folder on mobile now shows the folder's name at the top.
Deleting a custom design now asks for confirmation, so a single tap no longer removes a saved look and its reference images.
Fixed the remove (×) button on Custom Designs reference images not being tappable on phones; Save / Back buttons now stack on mobile.

July 2, 2026

Voice input in chat

AI Chat has a new microphone button next to Send so you can speak your message instead of typing it.
While recording, the input shows a live waveform that moves as you talk.
Tap stop and your words are transcribed into the message box, ready to edit or send.

Mobile welcome setup for new sign-ups

New sign-ups on mobile now get a quick, phone-friendly setup instead of skipping the welcome tour.
Pick what you want to make: flashcards, an AI video lecture, a quiz, or a study guide.
Snap a photo of your notes, choose a file, or paste text, and Scholarly starts building it right away.

Refreshed Help Center guides

All 58 help articles were checked against the current product, and 37 were updated to match what you see.
Updated guides cover the new model picker, PDF page selection in every create flow, and typeset document themes.
Guides now also reflect AI Meeting Notes, Deep Research file deliverables, and current video modes and lengths.
Four new guides: Paste Text, Studying from YouTube Videos, Deleting & Restoring Content, and Account, Login & Password Recovery.
Removed outdated information from help and product pages, including a free trial that no longer exists and retired features.

Smaller changes

Scanned or image-only PDFs from Google Drive now generate flashcards instead of failing with a “blank or unreadable” error.
Clearer errors for expired Google Drive connections, empty files, and not-yet-transcribed videos, with your credit refunded right away.
Fixed some YouTube videos failing with a “no transcript” error when building an AI Video Lecture or Podcast.
Fixed the “billing settings” link in payment-failed and yearly-renewal emails, which pointed to a page that didn’t exist.

July 1, 2026

Paste text as a study source

Use the new Paste Text option on Home or the Paste text tab in the Add content dialog to add material without a file.
Pasted notes, lecture transcripts, or articles become a saved source you can study from.
The Upload Content tile and the Add content dialog now make it clear that YouTube links are supported.

Redesigned chat model picker and citations

The model picker shows a focused Recommended Models list, with everything else in a separate Other Models menu.
The Gemini logo now appears in full color in the model picker.
Citations render as clickable source badges on every chat model, including while the answer is still streaming.

Smaller changes

Reorganized Home create options: Study Guide now comes first, and AI Chat and a new Folder shortcut appear in the main row.
Moved AI Timeline, AI SOP, AI Flowchart, AI Lesson Plan, AI Outline, Image Studio, and Practice Exam into the Home More menu.
Folders can now be renamed in place from the folder page header, with an option to generate a title with AI.
Flashcards now work with scanned or photographed PDFs from Google Drive by reading their pages as images.
Fixed the Custom Designs create button staying silently disabled; it now tells you what's missing.
The design picker now shows upfront that custom designs are a Premium feature.
Fixed Claude Sonnet 5 erroring out when Thinking was turned on in chat.
A model's thinking summary now appears above its answer instead of below it.
Fixed the chat model picker not showing a hover highlight in dark mode.
Gave the Add content dialog more comfortable side spacing on mobile.
Fixed the Build Flashcards button appearing off-center in the practice exam “What would you like to study?” empty state.

November 7, 2026

Build 3D models with the best model

Page through documents right in chat

Get help from any page, no account needed

Smaller changes

November 6, 2026

Remake a video lecture in another style

"Use my script as written" reads every word

More natural voices in podcasts and videos

Smaller changes

November 5, 2026

Study materials that follow your request more closely

See and sign out your devices

Mark up any image

Smaller changes

November 4, 2026

Create and work by voice

A visual video style picker

Smaller changes

November 3, 2026

Six video styles, with Shorts for every one

Smaller changes

November 2, 2026

24 more languages across Scholarly

Smaller changes

November 1, 2026

Ultimate video mode

Smaller changes

October 31, 2026

Word, PowerPoint, and cloud files work like PDFs

Build, explore, and share 3D models

Every creator takes every source

Smaller changes

October 30, 2026

A Tasks page for everything in progress

Make more from your recordings

A cleaner, easier-to-read Scholarly

Smaller changes

October 29, 2026

Cast your podcast, and pick from 66 voices

Ask for changes from the start

Real pictures in chat, in a swipeable row

Smaller changes

October 28, 2026

Pin messages in chat

Chat tasks file into your project or folder

Ask for a change while it's being made

Smaller changes

October 27, 2026

Choose pages or chapters

A cleaner, more complete help center

Smaller changes

October 26, 2026

Italian and Turkish

Scholarly in Arabic

Standard Fast video lectures

Smaller changes

October 25, 2026

Chat answers with the best model

Marketing and help in your language

Smaller changes

October 24, 2026

Chat reads your whole source and shows where it found the answer

Creating from chat just works

Change what Scholarly made, right in chat

Smaller changes

October 23, 2026

Real images in chat

Listen to a lecture instead, or send it to your phone

See who finished a lecture shared with your team

Smaller changes

October 22, 2026

Scholarly in nine more languages

Chats that recover, documents that still get built

From your file to a video lecture, faster

Smaller changes

October 21, 2026

A clearer progress page while you wait

Smaller changes

October 20, 2026