Skip to main content
CRITICAL

CVE-2026-22778

CVSS v3

9.8

CRITICAL

EPSS Score

3.8 %

exploit probability, as of 2026-09-25

CISA KEV

No

known exploited

Exploitation

—

SSVC status

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.

Technical details

Published
2026-02-02
Last Modified
2026-02-23

Frequently asked questions

What is 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.

Is CVE-2026-22778 actively exploited?

Active exploitation of CVE-2026-22778 has not been confirmed. Its EPSS score was 3.8% on 2026-09-25, the estimated probability of exploitation in the next 30 days.

What is the CVSS score for CVE-2026-22778?

CVE-2026-22778 has a CVSS v3 base score of 9.8 (CRITICAL severity).

Is CVE-2026-22778 affecting your environment?

Use isMalicious to check if any of your IPs or domains are associated with this vulnerability's IOCs.

Check an IPSearch CVEsSIEM & TIP integrations

No credit card required · 500 free checks/month · Free API key

Other 2026 vulnerabilities worth triaging

Ranked by exploit probability (EPSS).

Back to CVE Database