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
- [Bug]: 0.28.0 and 0.29.0 consume all host memory at start and freeze (OK with 0.27.1)
- [ROCm/MoRIIO] Route READ-mode remote KV load via async lifecycle (fix scheduler req_id_to_index KeyError)
- [Bugfix] Honor safetensors index tensor assignments
- [Regression 0.27→0.28+]: GlmMoeDsa (GLM-5.3) + decode-context-parallel: crashes on 0.28.0, silently returns random tokens on 0.29.0
- [Examples] Add SynapticChain native HTTP 402 pay-per-token vLLM proxy
- [Benchmark] Add multilingual character throughput metrics to bench serve (#51963)
- [Bugfix] Keep the frontend profiler on one thread, export off it
- [Bugfix][Frontend] Copy caller's prompt list at streaming-session entry
- [Model] Mistral Llama 4 scaling: compute once per forward, reuse across layers
- [ROCm][Perf] Use the fused clamped SwiGLU kernel on ROCm
- Docs
- Python not yet supported