CVE-2025-62372
From version 0.5.5 to before 0.11.1, users can crash the vLLM engine serving multimodal models by passing multimodal embedding inputs with correct ndim but incorrect shape (e.g. hidden dimension is wrong), regardless of whether the model is intended to…
Does this matter?
High impact if exploited, but EPSS currently rates exploitation as unlikely (0.38%). Schedule it in the normal patch cycle and watch for a rise in EPSS or a public exploit.
Description
vLLM is an inference and serving engine for large language models (LLMs). From version 0.5.5 to before 0.11.1, users can crash the vLLM engine serving multimodal models by passing multimodal embedding inputs with correct ndim but incorrect shape (e.g. hidden dimension is wrong), regardless of whether the model is intended to support such inputs (as defined in the Supported Models page). This issue has been patched in version 0.11.1.
- CVSS 4.0
- 8.3 HIGHCVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:N/VI:N/VA:H/SC:N/SI:N/SA:H/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/M
- EPSS
- 0.38% probability · 32th percentile
- CISA KEV
- Not listed
- Weakness
- CWE-129
- Affected
- vllm/vllm
- Source
- security-advisories@github.com
References
- https://github.com/vllm-project/vllm/commit/58fab50d82838d5014f4a14d991fdb9352c9c84bPatch
- https://github.com/vllm-project/vllm/pull/27204Issue Tracking, Patch, Vendor Advisory
- https://github.com/vllm-project/vllm/pull/6613Issue Tracking
- https://github.com/vllm-project/vllm/security/advisories/GHSA-pmqf-x6x8-p7qwMitigation, Vendor Advisory
Source: NVD record, EPSS from FIRST.org, KEV from CISA. Refreshed daily.