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 supported51 Subscribers
View all SubscribersAdd a CodeTriage badge to vllm
Help out
- Issues
- [Bugfix] Eliminate frontend ZMQ port TOCTOU with inherited listeners
- [Bug]: Evaluating GLM-5.1-FP8 for reasoning over large contexts produces unexpectedly poor level of accuracy.
- [Bug]: DeepGEMM is reported supported on sm_121 (GB10) but faults — support_deep_gemm() accepts the whole 120 capability family
- [Kimi-K3] AG-GEMM for Sequence Parallelism
- [Spec Decode] Support sample_from_anchor for DFlash draft models
- [Bug] Qwen3.8-Flash-Next: CUBLAS_STATUS_INTERNAL_ERROR / illegal memory access in GDN path with prefix caching on GB10 (sm_121); --no-async-scheduling does not help
- [Bug]: Quark FSE compatibility check matches `shared_expert_gate`, silently disabling fused shared experts on Qwen MoE MXFP4
- [Bugfix][V1] Use the Mamba cache group's block size for align-mode chunk splitting
- [Bug]: DFlash2 + YaRN identical 1.04M prompt gets zero prefix-cache reuse while target-only reuses ~1.039M tokens
- [Model] Support Ling 3.0 D-Spark serving
- Docs
- Python not yet supported