Vulnerability Intelligence Report
vLLM leaks a heap address when PIL throws an error
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.
Nuclei Template
CVSS Base Score
9.8
CRITICAL
Exploitability:3.9
Impact Score:5.9
EPSS Probability:3.82%
Executive Threat Verdict
Evaluating...
Evaluating Threat Landscape...
Assessing known weaponization, exploitation telemetry, and federal advisories.
Attack Surface
—
Authentication
—
Weaponization
—
SSVC Action
—
Weaknesses (CWE)
CWE-532 ↗CWE-532: Insertion of Sensitive Information into Log File
Affected Products & Versions
| Vendor | Product | Affected Versions |
|---|---|---|
| vllm-project | vllm | >= 0.8.3, < 0.14.1 (affected) |
References & Technical Advisories
No reference links found.
Threat Intelligence Signals
Identity & Timeline
| Status | PUBLISHED |
| Assigning Authority | GitHub, Inc. · Vendor · USA |
| Reserved | 2026-01-09T18:27:19 |
| Published | 2026-02-02T21:09:53 |
| Last Updated | 2026-07-15T01:19:16 |
Community Chatter & Buzz