What is Kimi K2.6?

Kimi K2.6 is the latest model from Moonshot AI. It's a 1-trillion-parameter open model built on a Mixture-of-Experts architecture (32B parameters active per token), and it was designed from the ground up for long-horizon coding, multi-step reasoning, and multi-agent work. For students, that translates into one thing: it's really good at building things end-to-end without losing the plot.

Why Add It to Scholarly?

When you ask Scholarly to generate a slide deck or a video lecture, the AI behind the scenes has to do a lot at once — research your sources, plan the structure, write code to render visuals, validate output, and iterate. Kimi K2.6 was built for exactly this kind of loop. Early internal testing showed cleaner slide layouts, more coherent video narrative arcs, and fewer retries on complex decks.

Where You'll See It

Chat. Kimi K2.6 is now selectable in the model picker inside any chat. Look under Premium Models.

AI Slides. When you create a new slide deck, Kimi K2.6 shows up in the model selector alongside Grok, GPT, and Gemini. It's a strong pick for technical subjects with dense material — the 262k context window lets it hold long textbooks and research papers in working memory while it designs the deck.

AI Video Lectures. Same story here. Kimi K2.6 is selectable before you generate a lecture. It's particularly good when your source material spans multiple PDFs or covers a broad topic that requires deliberate scene sequencing.

The 262k Context Window

One of the most practical wins is context size. Kimi K2.6 has a 262,144-token context window — large enough to keep an entire textbook, several research papers, or a full semester's worth of notes resident in the model's working memory. That means fewer re-reads, fewer missed details, and better citations when the AI builds your study materials.

Routed Through Cloudflare

Under the hood, we route Kimi K2.6 through Cloudflare Workers AI. That gives Scholarly users in the US faster responses and consistent availability. If you've used other Moonshot-hosted endpoints, you may notice Scholarly's Kimi feels snappier — that's why.

Pricing

Kimi K2.6 is available on paid plans, alongside our other premium models (GPT 5.4, Gemini 3.1 Pro, Grok 4.3). If you're already on Plus or Ultimate, you have access today — no upgrade needed.

How to Try It

1. Open any chat, click the model selector, and pick Kimi K2.6 under Premium Models.
2. Or head to your library, start creating AI Slides or an AI Video Lecture, and pick Kimi K2.6 in the model dropdown before generating.

What's Next

We're continuing to fine-tune how each model is prompted on Scholarly. Kimi K2.6 ships with a model-specific system prompt tuned to its strengths — so do GPT, Gemini, and Grok. You pick the model; we handle the prompting. If you have feedback or want to share what's working, we're always listening.