Enable extended thinking on supported models via the Responses API.
Not yet available. Reasoning support in the Responses API is on the roadmap. The page below documents the intended behavior.
Some models — particularly those trained for complex problem-solving — can spend additional compute on an internal reasoning pass before producing their final answer. The Responses API exposes this through a reasoning parameter.
Pass a reasoning object in your request to activate extended thinking:
| Value | Behaviour |
|---|---|
"low" | Minimal reasoning pass; fastest and cheapest |
"medium" | Balanced — good for most tasks |
"high" | Maximum reasoning budget; best for hard math, code, and multi-step logic |
When reasoning is enabled, the response output array may include a reasoning item before the final message:
The reasoning item contains a summary of the model's internal deliberation. Not all models surface a full trace — some return only the summary.
Reasoning tokens are billed the same as output tokens. A high effort request on a capable model can generate several thousand reasoning tokens before the visible answer, so monitor your usage field. See Reasoning Tokens for cost guidance and controls.
Only models that support extended thinking honor the reasoning parameter. For models that do not, the field is silently ignored. Filter for reasoning-capable models via the catalog API: