Skip to content

Commit 1ebdd83

Browse files
committed
docs: distinguish response_format from Responses text.format
1 parent df4dbe7 commit 1ebdd83

1 file changed

Lines changed: 5 additions & 4 deletions

File tree

‎docs-site/src/content/docs/reference/proxy-formats.md‎

Lines changed: 5 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -139,10 +139,11 @@ Responses items; translates function tools, tool choice, images, reasoning effor
139139
response formats; runs the normal Responses routing pipeline; then translates the result back.
140140

141141
Structured output is part of that translation: `response_format` with `json_object` or
142-
`json_schema` is forwarded to routed `openai-chat` models as-is, and rides the raw request body
143-
on native Responses routes. A backend without structured-output support returns its own error
144-
instead of the proxy rejecting the request locally. The same applies to `text.format` on
145-
`POST /v1/responses` when the model routes to an `openai-chat` provider.
142+
`json_schema` is forwarded to routed `openai-chat` models. On `POST /v1/responses` the
143+
equivalent request field is `text.format`: native Responses routes preserve it in the raw
144+
Responses body, and it is translated to `response_format` when the model routes to an
145+
`openai-chat` provider. A backend without structured-output support returns its own error
146+
instead of the proxy rejecting the request locally.
146147

147148
Non-streaming output has `object: "chat.completion"`. Streaming output uses SSE objects with
148149
`object: "chat.completion.chunk"`, choice deltas, a terminal choice with `finish_reason`, and

0 commit comments

Comments
 (0)