CVE-2025-62372

The record

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). The issue has existed ever since we added support for image embedding inputs, i.e. #6613 (released in v0.5.5)

Affected packages

Related vulnerabilities

• GHSA-pmqf-x6x8-p7qw vLLM vulnerable to DoS with incorrect shape of multimodal embedding inputs Same vulnerability
• CVE-2026-22778 vLLM is an inference and serving engine for large language models (LLMs). Also affects vllm
• GHSA-4r2x-xpjr-7cvv vLLM has RCE In Video Processing Also affects vllm
• GHSA-qh4c-xf7m-gxfc vLLM vulnerable to Server-Side Request Forgery (SSRF) through MediaConnector Also affects vllm
• CVE-2026-24779 vLLM vulnerable to Server-Side Request Forgery (SSRF) through MediaConnector Also affects vllm
• GHSA-2pc9-4j83-qjmr vLLM affected by RCE via auto_map dynamic module loading during model initialization Also affects vllm
• CVE-2026-22807 vLLM affected by RCE via auto_map dynamic module loading during model initialization Also affects vllm
• GHSA-grg2-63fw-f2qr vLLM is vulnerable to DoS in Idefics3 vision models via image payload with ambiguous… Also affects vllm