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
- [Feature] Skip MLA weight absorption when weights are pre-processed
- [Frontend] Support CustomToolCall in the non-Harmony Responses API
- [Frontend] Replay Responses reasoning via encrypted_content blobs
- [Frontend] Keep output_text logprobs on streamed Responses items
- [Feat][Mamba2] Enable internal prefill checkpoints
- [Bugfix][Frontend] Allow selecting Uvicorn HTTP implementation
- [Metrics] Add `vllm:kv_cache_token_usage_perc`
- [Model] Enable sequence-parallel RMSNorm for Qwen3 MoE
- [Refactor] Unify BlockStored event construction in cache_partial_block
- [Performance]: Canonical KV offload layout picks the DMA load path by page size, but copies per fragment
- Docs
- Python not yet supported