Studio: honor the Responses API text.format structured output - #8846
Studio: honor the Responses API text.format structured output#8846NilayYadav wants to merge 1 commit into
Conversation
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 72debc271a
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
|
@codex review |
|
Codex Review: Didn't find any major issues. Breezy! Reviewed commit: ℹ️ About Codex in GitHubCodex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
If Codex has suggestions, it will comment; otherwise it will react with 👍. When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback". |
1 participant
/v1/responses accepted text.format and dropped it: the field is typed on ResponsesRequest but _build_chat_request never read it, so a json_schema request returned free-form prose with HTTP 200. openai-python's responses.parse(text_format=...) and the Agents SDK's output_type= both fail on that, while the same schema through /v1/chat/completions response_format works.
Map text.format onto response_format (json_schema / json_object), which _extract_response_format already routes to llama-server's grammar. {"type": "text"} and a verbosity-only text stay unconstrained, so ordinary Codex / Agents SDK requests are unchanged. This is the inverse of the mapping external_provider already does when Studio is the client of a
Responses API.