---
id: CVE-2026-105758
title: vLLM is an inference and serving engine for large language models
summary: >-
  vLLM is an inference and serving engine for large language models. From 0.24.0
  until 0.30.0, the Qwen2VLVideoBackend and Qwen3VLVideoBackend classes accept
  request-level values for the media_io_kwargs.video.max_frames and
  media_io_kwargs…
severity: medium
cvss: 5.3
cvssVector: 'CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:L'
cwe:
  - CWE-770
vendor: vllm
product: vllm
affected:
  - 'vllm >= 0.24.0, < 0.30.0'
patched:
  - vllm 0.30.0
published: '2026-10-05'
updated: '2026-10-08'
sourceUpdated: '2026-10-08T03:16:35.407'
source: NVD
sourceUrl: 'https://nvd.nist.gov/vuln/detail/CVE-2026-105758'
references:
  - url: >-
      https://github.com/vllm-project/vllm/commit/ea723c81c3ea26425cb69503a5d5e90822a04a45
    label: security-advisories@github.com
  - url: 'https://github.com/vllm-project/vllm/pull/56729'
    label: security-advisories@github.com
  - url: 'https://github.com/vllm-project/vllm/releases/tag/v0.30.0'
    label: security-advisories@github.com
  - url: >-
      https://github.com/vllm-project/vllm/security/advisories/GHSA-x6mc-67gf-chw4
    label: security-advisories@github.com
  - url: 'https://github.com/advisories/GHSA-x6mc-67gf-chw4'
  - url: 'https://github.com/vllm-project/vllm'
  - url: >-
      https://security.access.redhat.com/data/csaf/v2/vex/2026/cve-2026-105758.json
  - url: 'https://access.redhat.com/security/cve/CVE-2026-105758'
  - url: 'https://bugzilla.redhat.com/show_bug.cgi?id=2546187'
  - url: 'https://www.cve.org/CVERecord?id=CVE-2026-105758'
  - url: 'https://nvd.nist.gov/vuln/detail/CVE-2026-105758'
  - url: 'https://pypi.org/project/vllm'
tags:
  - nvd
  - cve.org
  - ghsa
  - pip
  - osv
  - csaf
  - vex
  - red-hat
epss: 0.00303
epssPercentile: 0.22922
aliases:
  - GHSA-x6mc-67gf-chw4
  - PYSEC-2026-4310
ecosystem: pip
ssvc:
  exploitation: none
  automatable: 'yes'
  technicalImpact: partial
  timestamp: '2026-10-08T02:32:16.874942Z'
ingestedAt: '2026-10-05T23:36:21.157Z'
---

## Overview

vLLM is an inference and serving engine for large language models. From 0.24.0 until 0.30.0, the Qwen2VLVideoBackend and Qwen3VLVideoBackend classes accept request-level values for the media_io_kwargs.video.max_frames and media_io_kwargs.video.fps fields without enforcing server-side ceilings. An unauthenticated caller can submit these values to the /tokenize endpoint, causing the sampler to decode every frame selected from attacker-controlled video input, consume disproportionate frontend memory, and potentially terminate the API process before scheduling or admission control. The Rust frontend is not affected because it rejects the media_io_kwargs field. This issue is fixed in version 0.30.0.

## Affected

- `vllm >= 0.24.0, < 0.30.0`

## Remediation

Upgrade past the affected range:

- `vllm 0.30.0`

## Package advisory (CVE-2026-105758)

Affected packages:

- `vllm >= 0.24.0, < 0.30.0`

Patched in:

- `vllm 0.30.0`

Source: https://github.com/advisories/GHSA-x6mc-67gf-chw4

## Vendor advisories

- **Red Hat VEX** · Moderate · affected: Red Hat AI Inference Server, Red Hat Enterprise Linux AI (RHEL AI) 3, Red Hat OpenShift AI (RHOAI) · no fix planned: Red Hat AI Inference Server, Red Hat Enterprise Linux AI (RHEL AI) 3, Red Hat OpenShift AI (RHOAI) · updated 2026-10-06 · [vex](https://security.access.redhat.com/data/csaf/v2/vex/2026/cve-2026-105758.json)