fix(responses): generate Responses-compatible IDs in Chat Completions bridge

[Jwrede]

Relevant issues

Fixes #27333

Pre-Submission checklist

• I have Added testing in the tests/test_litellm/ directory, Adding at least 1 test is a hard requirement - see details
• My PR passes all unit tests on make test-unit
• My PR's scope is as isolated as possible, it only solves 1 specific problem
• I have requested a Greptile review by commenting @greptileai and received a Confidence Score of at least 4/5 before requesting a maintainer review

Type

• Bug Fix

Changes

The Chat Completions -> Responses API bridge reuses chatcmpl-* IDs from the underlying Chat Completion response as both the top-level response ID and the message output item ID. When bridged output from a non-OpenAI provider (e.g. Claude via LiteLLM) is later sent as input to an OpenAI Responses model, OpenAI rejects the request:

Invalid 'input[1].id': 'chatcmpl-dfa2da3a-...'. Expected an ID that begins with 'msg'.

This PR generates Responses-compatible IDs instead of reusing Chat Completions IDs:

• resp_<uuid> for the top-level ResponsesAPIResponse.id
• msg_<uuid> for message output items (GenericResponseOutputItem.id)
• img_<uuid> for image generation output items (OutputImageGenerationCall.id)

This matches what the streaming path already does (in streaming_iterator.py), which generates resp_ and msg_ prefixed IDs.

Files changed

• litellm/responses/litellm_completion_transformation/transformation.py -- generate resp_, msg_, img_ prefixed UUIDs instead of reusing chat_completion_response.id
• tests/test_litellm/responses/litellm_completion_transformation/test_response_id_prefixes.py -- 4 regression tests verifying correct ID prefixes on bridged responses
• tests/test_litellm/responses/litellm_completion_transformation/test_image_generation_output.py -- update existing test to assert img_ prefix instead of the old {id}_img_{idx} format

[Jwrede]

@greptileai

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[Jwrede]

@greptileai

[greptile-apps]

Greptile Summary

This PR fixes a bug where the Chat Completions → Responses API bridge reused chatcmpl-* IDs from the underlying provider response, causing OpenAI to reject bridged output when it was sent back as Responses API input. The fix generates proper resp_<uuid>, msg_<uuid>, and img_<uuid> prefixed IDs in the non-streaming path, aligning it with what the streaming iterator already does.

• transformation.py: Three targeted changes replace chat_completion_response.id reuse with fresh uuid.uuid4() calls using the correct prefixes for each output type.
• test_response_id_prefixes.py: Four new unit-only regression tests covering prefix correctness, ID distinctness, and dict-input handling.
• test_image_generation_output.py: Existing image ID assertions updated from the old {id}_img_{idx} exact-match to an img_ prefix check, accurately reflecting the new behaviour.

Confidence Score: 5/5

Safe to merge — the change is minimal, well-tested, and fixes a clear interoperability bug without touching auth, routing, or any critical request path.

The three code-site changes are straightforward and consistent with what the streaming path already does. New regression tests are mock-only and directly target the failure scenario described in the issue. The existing test update is appropriate: the old exact-ID assertion (test_123_img_0) would always fail with random UUIDs, and replacing it with a prefix check is the correct way to validate the new contract.

No files require special attention.

Important Files Changed

Filename	Overview
litellm/responses/litellm_completion_transformation/transformation.py	Replaces chatcmpl-* ID reuse with proper resp_/msg_/img_ UUID generation in the non-streaming Chat Completions → Responses bridge
tests/test_litellm/responses/litellm_completion_transformation/test_response_id_prefixes.py	New regression tests covering resp_/msg_ prefix enforcement and ID distinctness for bridged Responses output; pure unit tests using only mock data, no real network calls
tests/test_litellm/responses/litellm_completion_transformation/test_image_generation_output.py	Updates image ID assertions from exact chatcmpl-derived strings to img_* prefix checks, matching the new random-UUID generation behavior

_{Reviews (1): Last reviewed commit: "fix(responses): generate Responses-compa..." | Re-trigger Greptile}

[oss-pr-review-agent-shin]

🤖 litellm-agent: This PR was marked BLOCKED 7 days ago with no subsequent activity. Closing automatically.