vllm
https://github.com/vllm-project/vllm
Python
A high-throughput and memory-efficient inference and serving engine for LLMs
Triage Issues!
When you volunteer to triage issues, you'll receive an email each day with a link to an open issue that needs help in this project. You'll also receive instructions on how to triage issues.
Triage Docs!
Receive a documented method or class from your favorite GitHub repos in your inbox every day. If you're really pro, receive undocumented methods or classes and supercharge your commit history.
Python not yet supported59 Subscribers
View all SubscribersAdd a CodeTriage badge to vllm
Help out
- Issues
- [Bugfix][Mooncake] Settle decoder pull preparation failures
- [Quantization] Prefer Humming over Marlin for int4 g128 on SM12x
- [Model] Kimi K3.1 Pro (split 4/5: ROCm, P/D, indexer top-k)
- [ROCm][MiniMax-M3.1] MXFP8 linear on AITER's tuned FlyDSL GEMM
- [Bugfix] Fix MiniMax-M3 attention warmup deadlock under PP>1 (#60791)
- [Bug]: DSpark draft for DeepSeek-V4 models has no `packed_modules_mapping` → fused layers ignore `ignored_layers`, 0% acceptance (repro: DeepSeek-V4.1-Flash)
- [Bugfix][CPU] Support Mamba2 internal prefill checkpoints on CPU
- [Performance]: Batch-invariant mode on Hopper/Blackwell routes Linear layers to Triton although cuBLAS is already batch-invariant under the mode's settings
- [Bugfix][Frontend] Truncate singleton-batch prompt embeddings correctly
- [Model] Add Ling VL video, YaRN, and vision DP support
- Docs
- Python not yet supported