CVE-2026-22778
CRITICAL
GHSA-4r2x-xpjr-7cvv
CVSS Vector
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H
Lifecycle Timeline
3Tags
Description
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.
Analysis
Information exposure in vLLM inference engine versions 0.8.3 to before 0.14.1. Invalid image requests to the multimodal endpoint cause sensitive data logging. …
Sign in for full analysis, threat intelligence, and remediation guidance.
Remediation
Within 24 hours: Identify all systems running Vllm versions up to 0.14.1 and assess what sensitive data may have been logged. Within 7 days: Apply vendor patch to upgrade to Vllm 0.14.2 or later across all affected environments. …
Sign in for detailed remediation steps.
Priority Score
Vendor Status
Share
External POC / Exploit Code
Leaving vuln.today