CI / Build (push) Waiting to run
CI / Test (push) Waiting to run
When a reasoning-capable model has reasoning effort set to off, send 'none' rather than omitting the field or sending 'off'. This applies to both the ChatGPT Codex Responses API and OpenAI-compatible chat completions (both reasoning_effort and reasoning object formats). Non-reasoning models continue to send no reasoning field at all. Bumps version to 0.3.4.