Story
vllm_releases · Oct 6, 2026 · release
Source brief
vllm v0.31.0
github.comOct 6, 2026
original source linked
Release highlights
- DeepSeek-V4.1-Flash performance : FlashMLA mega attention with the V4.1 NVFP4 compressed KV cache is now the SM100 default ; DeepGEMM sparse MQA logits for t...
- Fast restart : the new vllm preload CLI launches the weight-cache daemon that keeps post-quantized weights resident in GPU memory across engine restarts , no...
- Model Runner V2 and speculative decoding : draft-model speculative decoding and custom logits processors on Model Runner V2; the new LiLiCorr drafter ; async...
Feed lens
agent