CVE-2026-22778

vLLM is an inference and serving engine for large language models (LLMs). From 0.8.3 to before 0.14.1, when an invalid image is sent to vLLM's multimodal endpoint, PIL throws an error. vLLM returns this error to the client, leaking a heap address. With this leak, we reduce ASLR from 4 billion guesses to ~8 guesses. This vulnerability can be chained a heap overflow with JPEG2000 decoder in OpenCV/FFmpeg to achieve remote code execution. This vulnerability is fixed in 0.14.1.

Scoring

Severity
CRITICAL
CVSS base score
9.8
CVSS vector
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H
EPSS probability
3.72%
CWE
CWE-532
Published
2026-02-02
Last modified
2026-07-15

Affected products

Weakness type

Related vulnerabilities

Markdown version · Browse all CVEs