MEDIUM 6.5 NVD
CVE-2026-115913
vLLM is an inference and serving engine for large language models. Prior to 0.32.0, the Phi-4 multimodal audio encoder in phi4mm_audio.py and phi4mm_utils.py us
vLLM is an inference and serving engine for large language models. Prior to 0.32.0, the Phi-4 multimodal audio encoder in phi4mm_audio.py and phi4mm_utils.py uses eager attention by default, allowing a remote client to submit one legal long audio clip that causes MultiHeadedAttention.forward_attention to materialize and retain full attention matrices in self.attn. The scaled dot-product attention branch does not materialize these matrices. The resulting quadratic GPU-memory use can cause an out-of-memory condition and terminate EngineCore for all users. This issue is fixed in version 0.32.0.
References
- https://github.com/vllm-project/vllm/commit/3d52578e13454de0bb8352cac988347cdcc89928
- https://github.com/vllm-project/vllm/pull/60614
- https://github.com/vllm-project/vllm/releases/tag/v0.32.0
- https://github.com/vllm-project/vllm/security/advisories/GHSA-8397-8qf6-f2gm
This medium severity vulnerability with a CVSS score of 6.5 was published on 2026-10-26 via NVD.
vulnfeed aggregates 18200 vulnerabilities from NVD, CISA KEV,
Ubuntu, Debian, Red Hat, Kubernetes, Exploit-DB, OSS-Security, GitHub and OpenStack — updated every 4 hours.