mirror of
https://github.com/BerriAI/litellm.git
synced 2026-10-11 03:38:38 +00:00
The Responses API delivers annotations as response.output_text.annotation.added events, and the Chat Completions stream bridge had no branch for them: a streamed reply never carried the url_citation that the same reply with stream=False reports on message.annotations (issue #43817). Emit each annotation on its delta exactly once, in Chat Completions format. The terminal response.completed payload repeats annotations, so deliberately not re-emitting them there (or on output_item.done) keeps accumulating clients from seeing duplicates. |
||
|---|---|---|
| .. | ||
| litellm_responses_transformation | ||
| __init__.py | ||
| test_litellm_responses_transformation_transformation.py | ||
| test_responses_bridge_provider_propagation.py | ||