Skip to content

fix(vllm): support developer chat messages#6077

Draft
skamenan7 wants to merge 2 commits into
ogx-ai:mainfrom
skamenan7:feat/6069-vllm-support-developer-message-type
Draft

fix(vllm): support developer chat messages#6077
skamenan7 wants to merge 2 commits into
ogx-ai:mainfrom
skamenan7:feat/6069-vllm-support-developer-message-type

Conversation

@skamenan7

Copy link
Copy Markdown
Collaborator

What does this PR do?

  • Converts vLLM chat completion developer messages to system messages before sending requests to the OpenAI-compatible vLLM endpoint, preserving content and name.
  • Keeps OGX accepting Codex-style developer messages while avoiding Qwen/vLLM chat template rejection.
  • Adds vLLM unit coverage for the outgoing message payload.

Related: #6069

Test Plan

  • Passed: uv run pytest tests/unit/providers/inference/test_remote_vllm.py -q (26 tests).

skamenan7 added 2 commits June 9, 2026 16:40
Signed-off-by: Sumanth Kamenani <skamenan@redhat.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant