CVE-2026-22773
vLLM is vulnerable to DoS in Idefics3 vision models via image payload with ambiguous dimensions
6.5
MEDIUM
CVSS 3.1
EPSS 0.40%
描述
vLLM is an inference and serving engine for large language models (LLMs). In versions from 0.6.4 to before 0.12.0, users can crash the vLLM engine serving multimodal models that use the Idefics3 vision model implementation by sending a specially crafted 1x1 pixel image. This causes a tensor dimension mismatch that results in an unhandled runtime error, leading to complete server termination. This issue has been patched in version 0.12.0.
如何修補 CVE-2026-22773
要修補 CVE-2026-22773,請將受影響套件升級到下列已修補版本。
- —升級至 0.12.0 或更新版本
- —升級至 0.12.0 或更新版本
CVE-2026-22773 正在被利用嗎?
低 — EPSS 為 0.4%,目前沒有觀察到大規模利用活動。
受影響套件(2)
- >= 0.6.4, < 0.12.0
- >= 0.6.4, < 0.12.0
CVSS 分數
| 來源 | 版本 | 嚴重程度 | 向量 |
|---|---|---|---|
| osv | CVSS 3.1 | MEDIUM6.5 | CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H |