Unverified50% confidenceReleaseExact time
vLLM released version v0.32.0rc3, which includes a bugfix for the Mamba2 chunk scan kernel producing NaN outputs on the NVIDIA SM90 architecture.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/17/2026
First Seen
Valid until: 10/31/2026
Sources
Related Entities
Related Claims
UnverifiedvLLM maintainer Bugen Zhao tagged the vllm-proto component with version proto-v0.4.0, corresponding to commit hash f28a508.74% similarUnverifiedvLLM released the v0.30.1rc0 pre-release version, introducing dense NVFP4 kernel and MoRI kernel image support for AMD MI355 accelerator cards (PR #58281).70% similarUnverifiedThe issue of the Mamba2 chunk scan kernel producing NaN on SM90 is very likely a numerical stability defect triggered by architectural differences.69% similarUnverifiedvLLM v0.27.0 enables the sm_107 target for the NVIDIA Rubin architecture and provides an NVLink all-reduce path on SM107.66% similarUnverifiedv3.0.68 switches GMI Cloud's default model to Qwen 3.8 Flash.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1049985API
curl https://kongchang.com/api/v1/knowledge/claims/1049985MCP
get_claim(id=1049985)