fix: Relax openai and pydantic pins to match vllm requirements (#9007) [r26.10]

[devin-ai-integration]

What does the PR do?

• Cherry-picks fix: Relax openai and pydantic pins to match vllm requirements #9007 (20e8b513) from main to r26.10 (clean pick, -x).
• python/openai/requirements.txt: openai==1.107.3 -> openai>=2.25.0,<3.
• src/python/pyproject.toml: pydantic==2.10.6 -> pydantic>=2.12.0,<3.
• Exact pins downgraded both packages in the vllm image, so tool_parsers failed to import NamespaceTool and the vllm model could not load.

Checklist

• PR title reflects the change and is of format <commit_type>: <Title>
• Changes are described in the pull request.
• Related issues are referenced.
• Populated github labels field
• Added test plan and verified test passes.
• Verified that the PR passes existing CI.
• Verified copyright is correct on all changed files.
• Added succinct git squash message before merging ref.
• All template sections are filled out.

Commit Type:

• fix

Related PRs:

• Original: fix: Relax openai and pydantic pins to match vllm requirements #9007
• triton-inference-server/vllm_backend r26.10 cherry-pick of fix: Support vllm>=0.30 bucket helper relocation in metrics vllm_backend#138

Where should the reviewer start?

python/openai/requirements.txt, src/python/pyproject.toml

Test plan:

• Same change as fix: Relax openai and pydantic pins to match vllm requirements #9007, validated on main by pipeline 72738572 (vLLM L0 jobs).

Caveats:

None; identical diff to #9007.

Related Issues:

• Resolves: TRI-2007

Link to Devin session: https://nvidia-cloud.devinenterprise.com/sessions/f4d32b335431447dbb84f0f9761a9110
Open in Devin Desktop: https://nvidia-cloud.devinenterprise.com/desktop/session/f4d32b335431447dbb84f0f9761a9110?variant=devin

[devin-ai-integration]

I'll fix CI failures and address comments from users with write access. I'll skip comments containing "(aside)".

• Disable automatic comment, CI, and merge conflict monitoring

[greptile-apps]

Confidence Score: 5/5

[High impact] The PR appears safe to merge; no actionable defect was established.

Summary

Replaces exact dependency pins with ranges intended to match vLLM's requirements.

• The OpenAI frontend can use OpenAI versions required by vLLM.
• The Triton Python package accepts newer Pydantic 2 versions.

_{Reviews (1) · Last reviewed commit: "fix: Relax openai and pydantic pins to m..." · Reviewed by Greptile}