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.
Fix
Patch, vendor advisory and mitigation references on NVD 3
- https://github.com/vllm-project/vllm/commit/3d52578e13454de0bb8352cac988347cdcc89928 Patch
- https://github.com/vllm-project/vllm/pull/60614 Patch
- https://github.com/vllm-project/vllm/security/advisories/GHSA-8397-8qf6-f2gm Exploit, Patch, Vendor Advisory
Every line above is quoted from NVD, CISA, Microsoft, the vendor's own advisory or DISA; nothing is generated.
Change history 6
- Cpes+1 CPE criteria (now 1)NVD modified 2026-10-28 15:38
- StatusPublished → AnalyzedNVD modified 2026-10-28 15:38
- StatusAwaiting Analysis → PublishedNVD modified 2026-10-26 16:56
- StatusPublished → Awaiting AnalysisNVD modified 2026-10-26 18:34
- StatusReceived → PublishedNVD modified 2026-10-26 16:56
- First seenCVSS 6.5NVD modified 2026-10-26 17:17
Affected (CPE criteria) 1
cpe:2.3:a:vllm:vllm:*:*:*:*:*:*:*:*
References 4
- https://github.com/vllm-project/vllm/commit/3d52578e13454de0bb8352cac988347cdcc89928 Patch
- https://github.com/vllm-project/vllm/pull/60614 Patch
- https://github.com/vllm-project/vllm/releases/tag/v0.32.0 Release Notes
- https://github.com/vllm-project/vllm/security/advisories/GHSA-8397-8qf6-f2gm Exploit, Patch, Vendor Advisory