vLLM Released Version v0.27.0 on 10 Aug 2026 with PyTorch 2.13 Upgrade and Kimi K3 Support
vLLM released version v0.27.0 on 10 Aug 2026 (tagged 4bdc8a7), featuring 561 commits from 242 contributors including 64 new ones. This release upgrades PyTorch…
Published on MyPrivateClaw
Aug 11, 2026, 5:26 PM UTC
Coverage date
Aug 10, 2026
Last updated
Aug 11, 2026, 5:26 PM UTC
News summary
vLLM released version v0.27.0 on 10 Aug 2026 (tagged 4bdc8a7), featuring 561 commits from 242 contributors including 64 new ones, released by project maintainer khluu. This release upgrades PyTorch to 2.13.0 along with torchvision 0.28.0 and Triton 3.7.1, explicitly marked as a breaking environment change; XPU and CPU backends also followed to torch 2.13. vLLM v0.27.0 adds full stack Kimi K3 support including core model files and kernels, Python and Rust frontends, AttnRes kernels, DeepGEMM support, compressed tensors quantized checkpoints, DSpark AR fusion, and optional shared expert sharding. Model Runner V2 expands to non generative workloads in this release, adding encoder only attention, sequence pooling for embedding/classification, encoder token classification and token embedding, BGE M3 pooling, multimodal on CPU, a multi layer MTP speculator, and PCP now selecting MRV2. FlashAt…