Codex App: add a None Intelligence/reasoning-effort option for supported models

Open 💬 0 comments Opened Apr 28, 2026 by uexoo

What variant of Codex are you using?

App

What feature would you like to see?

Please add a None option to the Codex App Intelligence/reasoning-effort selector for models that support no reasoning effort.

The API supports reasoning.effort = "none" for supported newer models, and this is useful for simple tasks where I want lower latency and fewer reasoning tokens, see OpenAI API Docs.

Other TUIs expose this option, but the Codex App UI does not currently let me choose None from the Intelligence selector.

Additional information

This would make the Codex App match the API more closely and give users an explicit low-latency / no-reasoning choice instead of forcing at least a reasoning-effort level.

View original on GitHub ↗