Repository navigation
[Doc] Update README and docs for v0.30.0 - #8175
Conversation
Announce the v0.30.0 release in the README news section (unified full-duplex framework, cross-stage KV/multimodal transfer, LingBot World interactive serving, MiniMax H3 Turbo on Blackwell/Ascend 950) and refresh the supported-model examples (Tencent AuK, Breeze-TTS-2, LingBot World, MAGI-2, pi0.5; drop removed dots.tts-era examples). Point the pre-built wheel and image instructions at the published 0.30.0 artifacts (PyPI wheel verified; ROCm 0.30.0/rocm723 wheel index verified), update the vLLM configuration and environment-variable references from 0.29 to 0.30, and move the NPU source-clone tag from v0.30.0rc1 to the final v0.30.0 release. Signed-off-by: hsliu_ustc <hsliu_ustc@noreply.gitcode.com>
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
|
This PR matches CODEOWNERS paths: /docs/. Code owners: @david6666666 @Gaohan123 Routing: @david6666666 via CODEOWNERS; @Gaohan123 via CODEOWNERS @hsliuustc0106, please review your own changes and leave a short self-review comment describing what you checked. PRs without author self-review may not be assigned a reviewer. Please take a look when you have a chance. If you would like an automated review, mention @vllm-omni-review-bot in a comment. |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: c88408d825
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| --env "HF_TOKEN=$HF_TOKEN" \ | ||
| -p 8091:8091 \ | ||
| vllm/vllm-omni-rocm:v0.28.0 \ | ||
| vllm/vllm-omni-rocm:v0.30.0 \ |
There was a problem hiding this comment.
For ROCm users following the recommended Docker setup, this newly substituted tag is not produced by the repository's release pipeline, so docker run will fail unless someone performs an undocumented out-of-band upload. I checked .buildkite/release/release-pipeline.yml: both release-image jobs build docker/Dockerfile.cuda, while .buildkite/release/scripts/publish-release-images.sh hardcodes vllm/vllm-omni; a repo-wide search found no publisher for vllm/vllm-omni-rocm. The same document also says availability of these prebuilt ROCm images is still tracked in issue #7405, so retain an actually published tag or add the ROCm publication step before advertising v0.30.0.
Useful? React with 👍 / 👎.
There was a problem hiding this comment.
@andyluo7 @FayeSpica @xuechendi PTAL for hardware specfic images
There was a problem hiding this comment.
Fixed in 77dab60. Verified the finding against the repo before fixing: .buildkite/release/release-pipeline.yml builds only docker/Dockerfile.cuda, publish-release-images.sh pushes only vllm/vllm-omni, and #7405 (open) tracks the out-of-band ROCm uploads — v0.28.0 is the latest tag confirmed live on Docker Hub (linux/amd64, digest 0ffd58d3c299).
The two docker run examples are back on v0.28.0, and the pre-built-images section now states that the prebuilt ROCm image is published separately from the release pipeline with newer-tag availability tracked in #7405, so the tag lagging the 0.30.0 wheels reads as intentional. The source-build path stays on the upstream vllm/vllm-openai-rocm:v0.30.0 base, which is published. Happy to flip the run examples to v0.30.0 as soon as that tag lands on the Hub.
Omni ReviewBot routing recordAssigned Direct under experiment |
vllm-omni-review-bot
left a comment
vllm-omni-review-bot
left a comment
There was a problem hiding this comment.
Omni ReviewBot review
PR description
This documentation-only change retargets the public install and configuration pages from the 0.28.0 / 0.30.0rc1 pins to the published 0.30.0 release. The root and docs landing pages add a September 2026 news item and refresh the supported-model examples to names already listed in the in-tree recipe catalog. Users following pip, git-clone, or Docker commands are now pointed at 0.30.0 artifacts; the ROCm official-image commands still sit beside an open published-tag caveat.
Change flow
flowchart TD
Release["[EXISTING] GitHub v0.30.0 release"]:::existing --> News["[CHANGED] README.md and docs/README.md<br/>news plus model-example lists"]:::changed
Release --> Install["[CHANGED] Installation CUDA/ROCm/NPU docs<br/>drop 0.28.0 caveats; pin 0.30.0"]:::changed
Release --> Config["[CHANGED] Configuration docs<br/>vLLM 0.29 links to 0.30"]:::changed
News --> Readers["[EXISTING] Docs and README readers"]:::existing
Install --> Users["[EXISTING] pip, docker, and source clone"]:::existing
Config --> Users
classDef existing fill:#e5e7eb,stroke:#6b7280,color:#111827
classDef changed fill:#fef3c7,stroke:#d97706,color:#451a03,stroke-width:2px
classDef new fill:#dcfce7,stroke:#16a34a,color:#052e16,stroke-width:2px
classDef removed fill:#fee2e2,stroke:#dc2626,color:#450a0a,stroke-width:2px
CI at
c88408d8259d(2026-09-26T20:24:23.796419+00:00): required check(s) blocking:buildkite/vllm-omni(missing), and -5 more. Observed Buildkite:buildkite/omni-release(passed), and -2 more.
Findings
- [P1] Do not advertise vllm/vllm-omni-rocm:v0.30.0 while #7405 still tracks unpublished tags —
docs/getting_started/installation/gpu/rocm.inc.md:168
The same ROCm page still says prebuiltvllm/vllm-omni-rocmpublished-tag availability is tracked in #7405 (lines 93-94), then the recommended official-image path tells users todocker run ... vllm/vllm-omni-rocm:v0.30.0(line 168, repeated at 189). This PR dropped the "until artifacts are published" caveats elsewhere because 0.30.0 is out, but it does not establish that the ROCm Hub tag exists: the author noted Docker Hub was unreachable and inferred tags from the release workflow. A user who follows the official ROCmdocker runwill fail on a missing tag, while the source-build path that usesvllm/vllm-openai-rocm:v0.30.0remains the documented fallback. Keep the source-build examples, and do not advertisevllm/vllm-omni-rocm:v0.30.0until that tag is published (or add the ROCm publish step and drop the #7405 caveat).
Existing thread: #8175 (comment)
Omni ReviewBot: finding feedback[p1] Do not advertise vllm/vllm-omni-rocm:v0.30.0 while #7405 still tracks unpublished tags — The same ROCm page still says prebuilt This finding appears in the bot's COMMENT review, but GitHub could not place it as an inline diff comment. If you are the PR author and disagree, react 👎 to this comment. The disagreement will be shown to the maintainer; it does not approve or merge the PR. |
The release pipeline builds and publishes only the CUDA image (.buildkite/release/release-pipeline.yml builds docker/Dockerfile.cuda; publish-release-images.sh targets vllm/vllm-omni). The prebuilt ROCm image is uploaded out-of-band, and v0.28.0 is the latest tag confirmed live on Docker Hub (#7405). Revert the two docker run examples to v0.28.0 and state why the tag lags the 0.30.0 wheels, so users do not hit a missing-tag pull failure. Source-build examples stay on the upstream vllm/vllm-openai-rocm:v0.30.0 base, which is published. Signed-off-by: hsliu_ustc <hsliu_ustc@noreply.gitcode.com>
|
Self-review (author):
|
2 participants
Follows the pattern of #6858 (v0.28.0 doc update) for the v0.30.0 release.
Documentation
2026/09news bullet for 0.30.0 (unified full-duplex framework around engine-owned sessions for MiniCPM-o 4.5/AURA, cross-stage KV and multimodal payload transfer via Mooncake/NIXL, LingBot World interactive world-model serving, realtime MiniMax H3 Turbo on Blackwell/Ascend 950) and refreshed the supported-model examples: Tencent AuK and Breeze-TTS-2 under TTS, LingBot World and MAGI-2 under diffusion, π0.5 under robot-policy.vllm/vllm-omni:v0.30.0,vllm/vllm-omni-rocm:v0.30.0).v0.30.0rc1to the finalv0.30.0tag.Verified: the 0.30.0 wheel is live on PyPI,
wheels.vllm.ai/rocm/0.30.0/rocm723returns HTTP 200, both newdocs.vllm.ai/en/v0.30.0links return HTTP 200, and the recipe links in the news bullet resolve in-tree. Docker Hub was unreachable from the review machine, so thev0.30.0image tags are asserted from the release workflow that publishes them alongside the GitHub release (same as #6858 did on its release day); happy to re-verify once the network allows.cc @hsliuustc0106