Subagent model_reasoning_effort config is ignored on Codex App
What version of the Codex App are you using (From “About Codex” dialog)?
Version 26.608.12217 • Released Jun 9, 2026
What subscription do you have?
Pro
What platform is your computer?
Microsoft Windows NT 10.0.26200.0 x64
What issue are you seeing?
After recent Codex App update, whenever an subagent I defined under ~\.codex\agents is spawned, if I work in Codex App, the subagent spawned will always be set with "medium" reasoning effort.
I tried in VS Code Codex Extension, it works normally. It can spawn subagents in configured model_reasoning_effort instead.
What steps can reproduce the bug?
Define whatever subagent, set its model_reasoning_effort to "high" or "xhigh".
Then ask Codex App to spawn the defined subagent.
You will see the subagent spawned with "medium" reasoning no matter how you set the model_reasoning_effort subagent config.
What is the expected behavior?
The model_reasoning_effort inside subagent config should be followed when that subagent is spawned.
Additional information
Only problem in Codex App. VS Code Extension works fine.
The Codex App also no longer directly show the reasoning effort at bottom statusline when opening subagent. I need to click /reasoning to see the reasoning in the subagent thread.
This issue has 3 comments on GitHub. Read the full discussion on GitHub ↗