Allow compaction to use a different model

Resolved 💬 2 comments Opened May 4, 2026 by franchesoni Closed Jul 1, 2026

What variant of Codex are you using?

CLI

What feature would you like to see?

I'd like to be able to configure codex so that if context exceeds 10% it will /compact but using a cheap model (like gpt-5.4-mini low). Is this possible? I think this could make users be more efficient

Additional information

_No response_

View original on GitHub ↗

This issue has 2 comments on GitHub. Read the full discussion on GitHub ↗