Package
vllm-openai-cuda-12.9
Component
vllm
Latest update
Fixed version
0.28.0-r0
Build, ship, and run secure software with minimal, hardened container images — rebuilt from source daily and guarded under our industry-leading remediation SLA.
Start for freeStatus
Fixed version
0.28.0-r0Status
Impact
GHSA-7h4p-rffg-7823 (CVE-2026-54235) is in vllm itself; the fix is in vllm 0.24.0. vllm-openai-cuda-12.9 is pinned to 0.19.1, the last release built for CUDA 12.9 (upstream moved its default CUDA to 13.0 at v0.20.0), and there is no 0.19.x maintenance release. Backporting needs a patch-policy exception, so this stays open until an upstream release compatible with CUDA 12.9 ships the fix. Customers needing the fix now should use vllm-openai-cuda-13.0 (vllm 0.28.0).
Status