Advisory

PYSEC-2026-143

Also known as: CVE-2026-22773, GHSA-grg2-63fw-f2qr

Summary

The record has no summary.

Details

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.

Severity

High Basis: CVSS v3 score 7.5, vector CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H

Dates

Published
2026-01-10 07:16 UTC
Modified
2026-05-20 09:19 UTC

Affected packages and fixes

vllm

PyPI · All systems

Affected

  • introduced 0.6.4, fixed 0.12.0 (ecosystem)

Fixed versions

0.12.0

Fix

  • Fixed version 0.12.0
    pip install --upgrade vllm==0.12.0

Fix commands are the export's own generic commands. Ojuju never runs them. Check them against your setup first.

Check your own computer: guard packages check

View this record on OSV.dev

Source

Source: PyPA Advisory Database · CC BY 4.0

Sources & licences