vllm-project/vllm
A high-throughput and memory-efficient inference and serving engine for LLMs
108 best points · 43 comments · 15 Hacker News posts · 96,493 GitHub stars · View on GitHub
Hacker News discussions
- vLLM v0.28.0 (108 points, 43 comments, by mrrrcs, 2026-08-29)
- vLLM introduces memory optimizations for long-context inference (5 points, 0 comments, by addisud, 2026-04-04)
- vLLM IR: A Functional Intermediate Representation for vLLM (4 points, 0 comments, by matt_d, 2026-04-07)
- Vllm (3 points, 0 comments, by kordlessagain, 2023-08-06)
- Vllm: High-throughput and memory-efficient inference and serving engine for LLMs (3 points, 0 comments, by tosh, 2023-09-10)
- vLLM 0.29.0: Model Runner V2 is now the default (3 points, 0 comments, by joshcsimmons, 2026-09-09)
- VLLM (2 points, 0 comments, by sherlockxu, 2023-06-25)
- Official PR Reveals the Inference Code for Mixtral 8x7B (2 points, 0 comments, by georgehill, 2023-12-11)
- Easy, fast, and cheap LLM serving for everyone (2 points, 0 comments, by vincent_s, 2023-12-17)
- vLLM (high-throughput LLM serving engine) (2 points, 0 comments, by roody_wurlitzer, 2026-02-25)
- Mixtral Expert Parallelism (1 points, 0 comments, by tosh, 2023-12-15)
- vllm (1 points, 0 comments, by tosh, 2023-12-15)
- VLLM Sacrifices Accuracy for Speed (1 points, 0 comments, by behnamoh, 2024-01-24)
- vLLM multi-turn conversations design (1 points, 0 comments, by CCs, 2026-01-22)
- vLLM: Jev-like mode for the DiffusionGemma model (1 points, 0 comments, by mmastrac, 2026-09-16)