Where does the bug appear (feature/product)?
Cursor IDE
Describe the Bug
I selected Composer 2 to run the prompt and I have a rule to return the model used in the prompt, it was showing ‘Composer 2’ correctly but it suddenly started showing Auto instead.
Steps to Reproduce
Run a prompt with Composer 2 with a rule set that it will return the name of the model its using for that prompt.
Expected Behavior
It should say Composer 2 was used.
Screenshots / Screen Recordings
Operating System
Windows 10/11
Version Information
Version: 2.7.0-pre.172.patch.0 (user setup)
VSCode Version: 1.105.1
Commit: de9b12bf051590906c465a8a9a221787ec3b88c0
Date: 2026-03-31T05:49:04.779Z
Layout: glass
Build Type: Stable
Release Track: Default
Electron: 39.8.1
Chromium: 142.0.7444.265
Node.js: 22.22.1
V8: 14.2.231.22-electron.0
OS: Windows_NT x64 10.0.26200
For AI issues: which model did you use?
Composer 2
Does this stop you from using Cursor
No - Cursor works, but with this issue
Hi @JaskiratOberoi,
The model’s self-reported name (when asked via a rule) isn’t a reliable way to verify which model is being used. Internally, it assigns a simplified identity label – either “Composer” or “Auto” – and there’s no “Composer 2” identity. Under certain routing paths, the label can say “Auto” even though Composer 2 is genuinely handling the request.
Your model routing is working correctly – Composer 2 is being used as you selected. The “I’m Auto” response is just the model following its identity instruction in the system prompt, not an indication that a different model was used.
Identiy instructions specify in cursor rules to inform me about the model used after each prompt. I tried it on my Mac to test again and it should me GPT 5.1 and called it Cursor’s agent, instead of Composer!
Version: 3.0.12
VSCode Version: 1.105.1
Commit: a80ff7dfcaa45d7750f6e30be457261379c29b00
Date: 2026-04-04T00:13:18.452Z
Layout: editor
Build Type: Stable
Release Track: Early Access
Electron: 39.8.1
Chromium: 142.0.7444.265
Node.js: 22.22.1
V8: 14.2.231.22-electron.0
OS: Darwin arm64 25.4.0
This actually reinforces the same point. The model is now saying “GPT-5.1” – which is also not the model being used. Each time you ask, you get a different answer (“Composer 2”, “Auto”, “GPT-5.1”) because models don’t reliably know their own product name. They guess based on training data, follow whatever the system prompt says, or fabricate a name entirely.
The name the model reports has no correlation with which model is actually handling your request. Composer 2 Fast is being used as you selected – the model just can’t accurately tell you that.
