Update qwen3.5-bf16-mi325x-sglang SGLang image to v0.5.12-rocm720-mi30x

[Klaud-Cold]

Updates SGLang image for qwen3.5-bf16-mi325x-sglang from v0.5.10-rocm720-mi30x to v0.5.12-rocm720-mi30x.
\nRef #1154

Generated with Claude Code

[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]

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=25980024237
see unofficial run visualizer at https://inferencex.semianalysis.com/evaluation?unofficialRun=25980024237

[claude]

🟡 The new perf-changelog.yaml entry at line 2556 has pr-link: XXX instead of an actual PR URL. All other entries in this file use the format https://github.com/SemiAnalysisAI/InferenceX/pull/<num>, so this should be replaced with https://github.com/SemiAnalysisAI/InferenceX/pull/1429 before merging.

Extended reasoning...

What the bug is

The diff in perf-changelog.yaml adds a new entry whose pr-link is the literal string XXX:

- config-keys:
    - qwen3.5-bf16-mi325x-sglang
  description:
    - "Update SGLang image from v0.5.10-rocm720-mi30x to v0.5.12-rocm720-mi30x"
  pr-link: XXX

XXX is the documented placeholder that authors use before a PR number is known. Per .github/workflows/claude.yml:224: "Use XXX as the PR-link placeholder until the PR exists." Once the PR exists, the placeholder is supposed to be replaced with the real URL.

Why the existing convention isn't followed

Of the 305 pr-link entries in perf-changelog.yaml, only 2 use XXX: one pre-existing entry at line 2502 and the new entry at line 2556 added by this PR. The 303 other entries — including every recently-added one (lines 2532, 2538, 2544, 2550) — use the full https://github.com/SemiAnalysisAI/InferenceX/pull/<num> URL.

Step-by-step proof

1. Open perf-changelog.yaml and scroll to line 2556.
2. Observe the value is the literal string XXX rather than a URL.
3. Compare against the immediately preceding entry at line 2554, which uses https://github.com/SemiAnalysisAI/InferenceX/pull/1402.
4. The current PR is Update qwen3.5-bf16-mi325x-sglang SGLang image to v0.5.12-rocm720-mi30x #1429, so the correct value is https://github.com/SemiAnalysisAI/InferenceX/pull/1429.

Impact

This is purely cosmetic — the Pydantic validator only requires pr-link to be a non-empty string, so nothing breaks at runtime. The harm is that the cross-reference back to the merging PR is lost, which makes perf-changelog.yaml less useful as a historical record (the whole point of including pr-link in each entry).

How to fix

Replace pr-link: XXX on line 2556 with:

  pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/1429

[github-actions]

see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=25980024338
see unofficial run visualizer at https://inferencex.semianalysis.com/evaluation?unofficialRun=25980024338

[github-actions]

see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=25980024237
see unofficial run visualizer at https://inferencex.semianalysis.com/evaluation?unofficialRun=25980024237

[github-actions]

see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=25980024338
see unofficial run visualizer at https://inferencex.semianalysis.com/evaluation?unofficialRun=25980024338

[github-actions]

see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=25980024237
see unofficial run visualizer at https://inferencex.semianalysis.com/evaluation?unofficialRun=25980024237

[github-actions]

see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=25984518223
see unofficial run visualizer at https://inferencex.semianalysis.com/evaluation?unofficialRun=25984518223

[functionstackx]

/reuse-sweep-run

[github-actions]

see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=25985608494
see unofficial run visualizer at https://inferencex.semianalysis.com/evaluation?unofficialRun=25985608494