OpenAI introduced GPT 5.5 on April 23 and made it available in the API on April 24. Scholarly now supports it in chat as the recommended premium OpenAI model when you want the strongest reasoning path.

GPT 5.5 is a bigger, more capable model than GPT 5.4. It is also substantially more expensive to run: OpenAI's standard API pricing is $5.00 per 1M input tokens, $0.50 per 1M cached input tokens, and $30.00 per 1M output tokens. Scholarly plan pricing is unchanged.

What's New in GPT 5.5

GPT 5.5 is built for complex professional work. Compared to GPT 5.4, you should notice:

• Stronger multi-step reasoning for coding, research, and detailed study questions
• Better long-context work when you bring in large documents or multi-part tasks
• More reliable tool use for workflows that need the assistant to create, inspect, and revise outputs
• Cleaner long-form structure in essays, study guides, and technical explanations

The Updated OpenAI Lineup

Scholarly now uses the OpenAI lineup that is actually available in the API:

• GPT 5.5 — The recommended premium OpenAI chat model. Best when you need careful reasoning, coding help, research, or long structured answers.
• GPT 5.4 Mini — The free default and lightweight generation model. It continues to power fast everyday chat, AI Slides, and AI Video Lectures.
• GPT 5.4 Nano — The low-cost internal model for short, fast tasks where speed matters most.

There is no GPT 5.5 Mini or GPT 5.5 Nano API model today, so we are not inventing those options in Scholarly.

What This Changes

AI Chat

• Paid users can use GPT 5.5 as the recommended premium OpenAI model.
• Free users continue to default to GPT 5.4 Mini so everyday chat stays fast and efficient.
• The Thinking toggle still works with GPT 5.5. It is the best OpenAI option when you want deeper reasoning.

AI Slides and Video Lectures

• AI Slides and AI Video Lectures continue to use GPT 5.4 Mini for the OpenAI lightweight path.
• That keeps long generations cost-efficient while still letting you choose other available models like Gemini, Grok, and Kimi where supported.

What About the Other Models?

The rest of the model lineup is unchanged. Gemini 3.1 Pro, Gemini 3 Flash, Grok 4.3, Grok 4.1 Fast, Kimi K2.6, and Claude remain available where they are supported. Different models still have different strengths, so we recommend GPT 5.5 for premium OpenAI reasoning and GPT 5.4 Mini when you want fast, efficient everyday answers.

For more on when to pick which, see our Choosing an AI Model guide.

Getting Started

There is nothing to install and nothing to migrate. Open chat, choose GPT 5.5 when you want the strongest OpenAI model, or keep GPT 5.4 Mini when speed and efficiency matter more.

---

Have feedback on GPT 5.5 in Scholarly? We want to hear about it. Email us at [email protected] — especially if you find a prompt where 5.5 does worse than 5.4 did. We watch every report.