Where does the bug appear (feature/product)?
Cursor IDE
Describe the Bug
I have a subagent that should be using Opus 4.6 but it appears to be using Composer 2 instead.
Steps to Reproduce
1 - Set up subagent with specified model. I’ve tried Opus and GPT Spark
2 - Trigger subagent from main agent chat
3 - Subagent runs. Composer 2 listed at the top
4 - I ask it what model is running, it says Composer. The subagent tab also says Composer. The performance also does not feel like Opus when running computer use tasks in the subagent as compared to the main agent.
Expected Behavior
I expect the Subagent’s model to respect the choice listed in the subagent. The performance feels degraded and when asking the subagent who it is, it claims it is Composer.
I provide the following in my subagent:
ame: sim-tap-opus
model: claude-4.6-opus-high-thinking
description: <removed>
---
You are a precision iOS simulator interaction agent.
<removed>
### For TAP objectives
Execute these steps in exact order. Never skip a step.
** Step 0 -- Declare yourself **
Declare what model you are. e.g. Composer, Opus, GPT, etc and which version too
See screenshot for what the agent claims.
Screenshots / Screen Recordings
Operating System
MacOS
Version Information
Version: 2.6.20 (Universal)
VSCode Version: 1.105.1
Commit: b29eb4ee5f9f6d1cb2afbc09070198d3ea6ad760
Date: 2026-03-17T01:50:02.404Z (4 days ago)
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: Darwin arm64 25.3.0
For AI issues: which model did you use?
Opus 4.6 high to trigger, Attempting to use same model in subagent.
For AI issues: add Request ID with privacy disabled
Always gives me “No request ID found”
Additional Information
This has been reported and mentioned to be fixed several times in Feb/Jan, but I am still having this issue
Does this stop you from using Cursor
Kind of, subagents are largely unusable.