Show Actual Model Used When “Auto” is Selected

[sultanmyrza]

Feature request for product/service

Chat

Describe the request

Currently, when I select Auto in Cursor Chat, the IDE decides which model to use for my code edits and responses. However, the chat header only shows “Auto”, so I cannot see which specific model (e.g., GPT-4, Claude, etc.) was actually used to generate the response.

Request:
Please update the UI so that when Auto is selected, Cursor displays the actual model chosen for that particular response. Possible options:
• Show the resolved model name next to “Auto” (e.g., Auto → GPT-4).
• Display the model used inline with the response metadata.
• Provide a tooltip or log entry with the model info.

Benefit:
• Improves transparency when Auto decides between models.
• Helps users evaluate output quality and consistency.
• Makes debugging and learning easier since we know which model handled the task.

[MidnightOak]

This has been asked for 1000 times. They aren’t going to.

[sultanmyrza]

Thank you for hint @MidnightOak by the way where can I upvote original feature request? Do you have any links?

[MidnightOak]

There is a couple related feature requests, but most of the chatter about identifying what model Auto is has been in discussions, not exactly a feature request.

Doesn’t hurt to have this feature request too, but Cursor has gone out of their way to make sure we can’t really identify what Auto is using. The model even used to say it could not answer questions about its origin/provider.

If the AUTO mode is enabled, I want that the response shows what Model used

When I choose auto mode, I hope that there is a label what Model it used somehow.

Auto Model Selection

When using Auto select for models it would be great if we could choose the models from which we auto select from to have preference for a selection of models. For example using Gemini-Flash and Gemini-Pro together as a small team for auto model selection. Having custom teams for auto model selection that could also work with API’s would be really cool as I tend to have prefences for model outputs for particular tasks. Further to this to have a group of models from which the auto selects and th…

Configurable Auto Model Selection

It would be great to have the ability to configure auto-selection of models based on task type. For example: Use a lightweight LLM for tooling commands and simple tasks. Then auto switch back to a stronger model (e.g., Claude 4 Sonnet) for coding. Assign a specific model for different categories of work e.g. bug fixing or other specialized tasks. This would save time, optimize costs, and let users leverage the right model for the right job without manually switching each time.