Repository navigation
[Recipe] MiniMax-M3: add B200 NVFP4 AgentX aggregated profile - #63163
Merged
wzhao18 merged 3 commits intoOct 23, 2026
Merged
Conversation
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> Signed-off-by: hshrivastava-droid <hshrivastava@nvidia.com>
![claude[bot]](https://avatars.githubusercontent.com/in/1236702?s=60&v=4){kind=small}
wzhao18
approved these changes
Oct 23, 2026
Contributor
|
/ci run |
wzhao18
enabled auto-merge (squash)
October 23, 2026 03:06
github-actions
Bot
added
the
ready
|
✅ Triggered Buildkite CI #96788 for commit |
github-actions Bot
added a commit
to vllm-project/recipes
that referenced
this pull request
Oct 23, 2026
Signed-off-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Contributor
|
/ci run |
|
✅ Triggered Buildkite CI #96904 for commit |
github-actions Bot
added a commit
to CuiZ2007/recipes
that referenced
this pull request
Oct 23, 2026
…mni@78610269b446 vllm-project/vllm@main:recipes c3117c7441dd → 37cf71255259 [latest.vllm] vllm-project/vllm@c3117c7...37cf712 Commits: - vllm-project/vllm@74ba6384c2f0 [Doc][ROCm] Document DeepSeek-V4.1-Flash MI355X AgentX TP2 profiles (vllm-project/vllm#63344) - vllm-project/vllm@ecd1e12553e6 [Recipe] MiniMax-M3: add B200 NVFP4 AgentX aggregated profile (vllm-project/vllm#63163) Recipes: - MiniMaxAI/MiniMax-M3 (modified): https://recipes.vllm.ai/MiniMaxAI/MiniMax-M3 - deepseek-ai/DeepSeek-V4.1-Flash (modified): https://recipes.vllm.ai/deepseek-ai/DeepSeek-V4.1-Flash vllm-project/vllm-omni@main:recipes c829af446829 → 78610269b446 [latest.vllm-omni] vllm-project/vllm-omni@c829af4...7861026 Commits: - vllm-project/vllm-omni@45d7c0051ec6 [Feat] Add Cosmos3-Nano Pipeline Parallel support (vllm-project/vllm-omni#4637) - vllm-project/vllm-omni@ea719bc6b99b [BugFix] security: restrict untrusted image decoding formats (vllm-project/vllm-omni#9114) Signed-off-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
github-actions Bot
added a commit
to shantanusharma/recipes
that referenced
this pull request
Oct 23, 2026
vllm-project/vllm@main:recipes 8fae0ff58bae → 7651825993d0 [latest.vllm] vllm-project/vllm@8fae0ff...7651825 Commits: - vllm-project/vllm@74ba6384c2f0 [Doc][ROCm] Document DeepSeek-V4.1-Flash MI355X AgentX TP2 profiles (vllm-project/vllm#63344) - vllm-project/vllm@ecd1e12553e6 [Recipe] MiniMax-M3: add B200 NVFP4 AgentX aggregated profile (vllm-project/vllm#63163) Recipes: - MiniMaxAI/MiniMax-M3 (modified): https://recipes.vllm.ai/MiniMaxAI/MiniMax-M3 - deepseek-ai/DeepSeek-V4.1-Flash (modified): https://recipes.vllm.ai/deepseek-ai/DeepSeek-V4.1-Flash Signed-off-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
This was referenced Oct 23, 2026
github-actions Bot
added a commit
to vllm-project/recipes
that referenced
this pull request
Oct 23, 2026
vllm-project/vllm@main:recipes c3117c7441dd → d561af5dfcd1 [latest.vllm] vllm-project/vllm@c3117c7...d561af5 Commits: - vllm-project/vllm@7105440c4009 [Doc] MiniMax-M3: use the autotune flag the pinned GB200 nightly accepts, fix golden-AL links (vllm-project/vllm#63394) - vllm-project/vllm@c7c11188a0a8 [Doc] MiniMax-M3: document the B300 NVFP4 InferenceX AgentX aggregated profile (vllm-project/vllm#63395) - vllm-project/vllm@8b21328cb844 [Doc] MiniMax-M3: document the H200 MXFP8 InferenceX AgentX aggregated profile (vllm-project/vllm#63391) - vllm-project/vllm@74ba6384c2f0 [Doc][ROCm] Document DeepSeek-V4.1-Flash MI355X AgentX TP2 profiles (vllm-project/vllm#63344) - vllm-project/vllm@ecd1e12553e6 [Recipe] MiniMax-M3: add B200 NVFP4 AgentX aggregated profile (vllm-project/vllm#63163) Recipes: - MiniMaxAI/MiniMax-M3 (modified): https://recipes.vllm.ai/MiniMaxAI/MiniMax-M3 - deepseek-ai/DeepSeek-V4.1-Flash (modified): https://recipes.vllm.ai/deepseek-ai/DeepSeek-V4.1-Flash Signed-off-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
bantianyue
pushed a commit
to bantianyue/vllm
that referenced
this pull request
Oct 28, 2026
…roject#63163) Signed-off-by: hshrivastava-droid <hshrivastava@nvidia.com> Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com> Co-authored-by: Wei Zhao <51183510+wzhao18@users.noreply.github.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Overview
Adds the B200 NVFP4 AgentX aggregated serving profile for MiniMax-M3 from SemiAnalysisAI/InferenceX#4382 (
minimaxm3-fp4-b200-vllm-agentic-mtpandminimaxm3-fp4-b200-vllm-agentic-mtp-c10).Claims
--max-num-batched-tokens 16384,--gpu-memory-utilization 0.95,--disable-flashinfer-autotune,--stream-interval 20, and the pinned B200 image.--max-num-seqs,--max-num-active-seqs, draft tokens, CUDA graph capture sizes, SimpleCPUOffload KV offload) for both InferenceX keys, with their digest-pinned images.Validation
node scripts/validate-source.mjs --source vllm --dir <vllm>/recipespasses against currentvllm-project/recipesmain.resolveCommandand compared it with the qualified command; TP4 requires overriding the builder's TP8 default, noted in the guide.Details
Per-point differences stay in the guide, following the GB200 AgentX section. The
-c10arms use a different pinned image,mnnvlall-reduce and no local argmax reduction, so they are documented separately. The builder's NVFP4 default setsminimax_m3_msa_decode_backend: cutlasson every point; the qualified points set it only for concurrency 30 and 38, which the guide states.claude-sonnet-5-5; prepared the recipe change and PR text).