Skip to main content

Streams an evaluator model's analysis comparing two previously generated A/B responses.

POST 
/studio/compare_results

Streams the evaluator model's analysis (as SSE) against the original prompt, then persists the structured verdict onto the parent OrcaRequest.

  • Validates the request id and loads the associated OrcaRequest with its tasks (A/B).
  • Constructs a system instruction plus user messages containing the original prompt and both responses.
  • Streams the evaluator's analysis with SSE headers; after streaming, persists the full analysis to the request.
  • Responds with appropriate HTTP codes for missing/invalid request or if not found.

Request​

Responses​

OK