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 supported44 Subscribers
View all SubscribersAdd a CodeTriage badge to vllm
Help out
- Issues
- [Kernels] Integration for DeepGEMM MegaMoE kernel
- [Spec Decode] Add block-wise verification rule to V1 rejection sampler
- [Feature]: DeepSeek-V4 Flash sm_80 (A100/A800) support
- [RFC]: Rust front-end
- [Bug]: vLLM TP=4 corrupts non-English output on PCIe consumer GPUs
- [Bug]: GDN-based models (Qwen3.5/3.6 family) crash on the second request when running on multi-GPU XPU with prefix caching enabled
- [Bug]: assert is_mixture_of_experts fails on vllm serve with --enable-eplb
- [Feature]: Support DDTree speculative decoding
- [Feature]: Make opencv-python-headless an optional dependency for FIPS compliance
- [Bug]: CUDA graph capture crashes during startup due to Inductor autotuning torch.cuda.synchronize() inside graph capture (FULL_DECODE_ONLY + MLA + FP8) when PDL is enabled
- Docs
- Python not yet supported