Model selection

Choose between three AI models optimized for speed, capability, and advanced reasoning.

Available models

Small model

GPT-4o mini — Small model for fast, lightweight tasks.

Large model

GPT-4o — Large model for complex, multi-step tasks.

Reasoning model

DeepSeek R1 — Uses advanced reasoning. Shows chain-of-thought output before the final answer.

When to use each model

Small model (GPT-4o mini)

Use the Small model when speed matters and the task is straightforward:

• Quick factual questions
• Summarizing short passages
• Simple code edits or completions
• Drafting short messages

This is the default model for all new conversations.

Large model (GPT-4o)

Use the Large model for demanding or multi-step work:

• Complex coding tasks involving multiple files or logic chains
• Long-form writing and editing
• Tasks requiring nuanced instruction-following
• Generating and iterating on artifacts

Reasoning model (DeepSeek R1)

Use the Reasoning model when the problem benefits from deliberate, step-by-step thinking:

• Math and logic problems
• Algorithmic planning
• Tasks where you want to inspect the model’s reasoning before accepting the answer

The model’s chain-of-thought is displayed as a collapsible Reasoning block above each response. See Reasoning model display for details.

How to switch models

The model selector is in the chat header, on the left side next to the chat title area.

1. Click the model selector button

   The button displays the name of the currently active model. Click it to open a dropdown menu.

2. Select a model

   Each option shows the model name and a short description. The currently active model has a filled checkmark on the right.

3. Selection takes effect immediately

   The choice is saved as a cookie and applied to the next message you send in any conversation.

You can switch models mid-conversation. Earlier messages are sent to the server with the model that was active when they were submitted; only new messages use the newly selected model.

Limitations of the Reasoning model

The Reasoning model does not support artifacts or tools. When the Reasoning model is selected, the createDocument and updateDocument tools are not available, and the AI will not open an artifact panel. Use the Small model or Large model if you need artifact functionality.

This restriction exists because the system prompt for the Reasoning model omits the artifacts instructions that are included for the other two models. If you start a conversation using the Reasoning model and then switch to the Small or Large model, artifact support becomes available from that point forward.