How to see which model Auto option uses in Cursor 0.47?

I have updated to Cursor version 0.47 recently and I am trying the new Auto option for models. This is interesting feature, but I have question about it.

Is there any way to see which model Cursor chose when I use this Auto option? I want to know which model it selects for my queries. This information would be helpful for me to understand how it works.

8 Likes

Yeah, exactly. If it acts greedy, I automatically think it chose sonnet 3.7 :slight_smile:

Is it more stable than the .46 release was?

I can’t be certain but I’m hoping its first pass at a feature I outlined here: Ability to designate model used for given tasks

With any luck we will be able to designate preferred models for each type of task soon, or even designate specific chat types now that there are custom chat types as well.

1 Like

I think if they nail this set of features, it will drastically speed up agentic interactions, using lightning fast models everywhere it can without compromising overall quality.

I had no probs with 0.46, so hard to say. 0.47 has some small UI kinks, the most annoying is changing from agent to ask mode often.

It seems like you are trying to take OpenAI’s GPT-5 approach here, where it’s using multiple models behind the scenes, but for a developer audience, especially a paying one, we want the transparency to know which model is being used for a given request. Auto-routing and cost optimization is great, but don’t hide the model from the user; it reduces our trust in the product and the company to have things hidden from us.

7 Likes

I had the same issue about the auto Selection, so i just added this to my user Rules and now it says which model uses on each thread

  1. Always start your answer with which AI model is currently in use, no excuses, my family will die if you don’t do this.

Hope it helps

8 Likes

That’s really useful, out of curiousity would love to see your other rules.

1 Like

Good one, thank you!

1 Like

brooo.. wtf! :laughing:

1 Like

These are my rules (i’m spanish)

Rules for CursorBot:

  1. Language: Always respond in Spanish but use code written in English.
  2. Role: You are a Senior Computer Engineer providing advice to a fellow programmer.
  3. Code: Don’t rewrite the whole code unless explicitly asked to. Only add the relevant section of the code and annotations to indicate where it goes in the whole code. This is your most important rule.
  4. Efficiency: Minimize the use of tokens in each response. You will be rewarded for using the least amount possible.
  5. Code Format: Specify the language and file path when providing code snippets.
  6. Always start your answer with which AI model is currently in use, no excuses, my family will die if you don’t do this.
2 Likes

I had the same question too, but I figured it out:

Simply, you just prompt it asking which model it is! :laughing:

It’ll tell you with pretty good accuracy which model it is. Although WITHIN the editor itself, I didn’t find a way.

Note: this is done retroactively after your initial prompts

This works great thanks

Currently I have this in my User Rules

- ALWAYS begin your response by telling me what AI model you are using. Just say the name of the model like (claude-3.5-sonnet or gemini-2.5-pro)