[Klaud Cold] Update dsv4-fp4-b300-vllm (+mtp) vLLM image to v0.22.0

[functionstackx]

Summary

Update vLLM image from v0.21.0 to v0.22.0

Recipes touched: dsv4-fp4-b300-vllm, dsv4-fp4-b300-vllm-mtp

Test plan

• full-sweep-enabled sweep passes.

🤖 Generated with Claude Code

---

Note

Low Risk
Only updates container image tags and changelog metadata for existing benchmark configs; no logic, auth, or sweep-parameter changes.

Overview
Bumps the vLLM OpenAI container for the DeepSeek-V4-Pro FP4 B300 single-node benchmark recipes dsv4-fp4-b300-vllm and dsv4-fp4-b300-vllm-mtp from v0.21.0 to v0.22.0 in .github/configs/nvidia-master.yaml. Sweep search spaces and other recipe fields are unchanged.

Documents the pin in perf-changelog.yaml for those config keys.

^{Reviewed by Cursor Bugbot for commit 14c6aae. Bugbot is set up for automated code reviews on this repo. Configure here.}

[github-actions]

Thanks for the contribution! For vLLM & SGLang, please ensure that your recipes is similar to the official vLLM recipes and/or the SGLang cookbook

• https://docs.sglang.io/cookbook/intro
• https://recipes.vllm.ai/

If it is not, please create a PR first before we can merge your single node PR into the master branch. Let's ensure that the documentation is first class such that the entire ML community can benefit from your hard work! Thank you

• https://github.com/vllm-project/recipes
• https://github.com/sgl-project/sglang/tree/main/docs_new

PR authors are responsible for ensuring that after merging, all GitHub Action jobs fully pass. A lot of the time, failures are just flakes and simply re-running the failed jobs will fix it. If re-running failed jobs is attempted, PR authors are responsible for ensuring it passes. See GitHub's docs on re-running failed jobs: https://docs.github.com/en/actions/how-tos/manage-workflow-runs/re-run-workflows-and-jobs#re-running-failed-jobs-in-a-workflow

As a rule of thumb, generally, PR authors should request a review & get a PR approval from the respective companies' CODEOWNERS before requesting a review from core maintainers.

If additional help is needed, PR authors can reach out to core maintainers over Slack.

[github-actions]

Thanks for the contribution! For vLLM & SGLang, please ensure that your recipes is similar to the official vLLM recipes and/or the SGLang cookbook

• https://docs.sglang.io/cookbook/intro
• https://recipes.vllm.ai/

If it is not, please create a PR first before we can merge your single node PR into the master branch. Let's ensure that the documentation is first class such that the entire ML community can benefit from your hard work! Thank you

• https://github.com/vllm-project/recipes
• https://github.com/sgl-project/sglang/tree/main/docs_new

PR authors are responsible for ensuring that after merging, all GitHub Action jobs fully pass. A lot of the time, failures are just flakes and simply re-running the failed jobs will fix it. If re-running failed jobs is attempted, PR authors are responsible for ensuring it passes. See GitHub's docs on re-running failed jobs: https://docs.github.com/en/actions/how-tos/manage-workflow-runs/re-run-workflows-and-jobs#re-running-failed-jobs-in-a-workflow

As a rule of thumb, generally, PR authors should request a review & get a PR approval from the respective companies' CODEOWNERS before requesting a review from core maintainers.

If additional help is needed, PR authors can reach out to core maintainers over Slack.

[github-actions]

see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=26696203941
see unofficial run visualizer at https://inferencex.semianalysis.com/evaluation?unofficialRun=26696203941

[claude]

LGTM — straightforward vLLM image version bump for two existing benchmark recipes.

Extended reasoning...

Overview

This PR bumps the vLLM OpenAI Docker image from v0.21.0 to v0.22.0 for the dsv4-fp4-b300-vllm and dsv4-fp4-b300-vllm-mtp benchmark recipes in .github/configs/nvidia-master.yaml, and adds a corresponding changelog entry in perf-changelog.yaml. No search spaces, scenarios, or other configuration are changed.

Security risks

None. This is a benchmark configuration change touching only Docker image tags and a changelog file — no application logic, authentication, cryptography, or permission paths are affected.

Level of scrutiny

Low. Config-only image pin changes for existing benchmark recipes are mechanical and follow the established pattern across many prior PRs in this repo. Correctness will be confirmed by the full-sweep sweep run gated by the full-sweep-enabled label.

Other factors

The change is self-contained, follows the exact pattern of other image-bump PRs in recent history (e.g. #1593, #1594), and the bug hunting system surfaced no issues. The changelog entry is properly linked to this PR.

[github-actions]

see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=26696210620
see unofficial run visualizer at https://inferencex.semianalysis.com/evaluation?unofficialRun=26696210620