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 supported40 Subscribers
View all SubscribersAdd a CodeTriage badge to vllm
Help out
- Issues
- [Performance] DP8 vs TP8 for single-KV-head MLA: 7.7x KV, 3.4x faster 1M TTFT at c=8 (DeepSeek-V4-Flash-0731, 8x B200, vLLM v0.25.0)
- [Performance] MiniMax-M3-NVFP4 on 8x B200, first numbers after the #48929 correctness fix: 1M real-prose envelope, EAGLE3 2.1-2.3x decode
- [Bugfix][Parser] Stream Inkling plain-text answers incrementally
- [Perf] Keep staged writes in NumPy chunks
- [Bugfix][Mooncake] Save exact Mamba boundary states
- Fix DCP max-size check to account for PCP-induced KV duplication
- [DeepEPv2] Support MXFp8 Activation Scale Dispatch
- [Model Loader][Perf] Auto-prefetch VirtioFS checkpoints
- [Online quantization] Add targeted online quantization configuration based on user patterns
- [Bugfix][Benchmark] Preserve SSE output across transport chunk boundaries
- Docs
- Python not yet supported