OpenAI Codex 5.2 Verbose thinking/reasoning

Has anyone else seen this behavior with Clawd Bot + OpenAI Codex 5.2?

When I set Codex 5.2 as the model, the bot still outputs reasoning content in iMessage, even though:
โ€ข verbose is off
โ€ข reasoning is explicitly set to off

Despite that configuration, internal reasoning is still being returned/displayed.

This does not happen with other models โ€” only with Codex 5.2.

Is this:
โ€ข a known issue with Codex 5.2?
โ€ข a Clawd Bot parsing / filtering issue?
โ€ข or expected behavior due to how Codex handles reasoning internally?

Any insights or workarounds would be appreciated.
Was this page helpful?