---
id: CVE-2026-114055
title: vLLM 0.7.3 before 0.31.0 DoS via Unvalidated mm_processor_kwargs
summary: >-
  vLLM from 0.7.3 before 0.31.0 contains an uncontrolled resource consumption
  vulnerability that allows remote clients to control multimodal processor
  parameters by supplying per-request mm_processor_kwargs. Attackers can set
  large image s…
severity: high
cvss: 7.5
cvssVector: 'CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H'
cvssSource: cna
cwe:
  - CWE-400
  - CWE-770
vendor: vllm-project
product: vllm
affected:
  - vllm >= 0.7.3 < 0.31.0
ssvc:
  exploitation: none
  automatable: 'yes'
  technicalImpact: partial
  timestamp: '2026-10-21T14:40:33.947720Z'
published: '2026-10-21'
updated: '2026-10-26'
sourceUpdated: '2026-10-26T19:49:33.622Z'
source: CVEORG
sourceUrl: 'https://www.cve.org/CVERecord?id=CVE-2026-114055'
references:
  - url: >-
      https://github.com/vllm-project/vllm/security/advisories/GHSA-gx7p-2j49-hfq4
    label: GitHub Security Advisory (GHSA-gx7p-2j49-hfq4)
  - url: >-
      https://www.vulncheck.com/advisories/vllm-0.7.3-before-0.31.0-dos-via-unvalidated-mm-processor-kwargs
    label: >-
      VulnCheck Advisory: vLLM 0.7.3 before 0.31.0 DoS via Unvalidated
      mm_processor_kwargs
  - url: 'https://nvd.nist.gov/vuln/detail/CVE-2026-114055'
  - url: 'https://github.com/vllm-project/vllm/pull/58830'
  - url: >-
      https://github.com/vllm-project/vllm/commit/31f2e70cd3e403aaa471fb7fe3d33a7207aaf3cc
  - url: 'https://github.com/vllm-project/vllm'
  - url: 'https://github.com/vllm-project/vllm/releases/tag/v0.31.0'
  - url: 'https://github.com/advisories/GHSA-gx7p-2j49-hfq4'
  - url: >-
      https://security.access.redhat.com/data/csaf/v2/vex/2026/cve-2026-114055.json
  - url: 'https://access.redhat.com/security/cve/CVE-2026-114055'
  - url: 'https://bugzilla.redhat.com/show_bug.cgi?id=2558025'
  - url: 'https://www.cve.org/CVERecord?id=CVE-2026-114055'
tags:
  - cve.org
  - osv
  - pip
  - ghsa
  - csaf
  - vex
  - red-hat
epss: 0.0042
epssPercentile: 0.36113
aliases:
  - GHSA-gx7p-2j49-hfq4
ecosystem: pip
patched:
  - vllm 0.31.0
ingestedAt: '2026-10-21T01:37:08.041Z'
---

## Overview

vLLM from 0.7.3 before 0.31.0 contains an uncontrolled resource consumption vulnerability that allows remote clients to control multimodal processor parameters by supplying per-request mm_processor_kwargs. Attackers can set large image size, patch count, crop count, or frame count values in Isaac, InternVL, H2OVL, and other processors to exhaust memory and terminate the shared server.

## Affected

- `vllm >= 0.7.3 < 0.31.0`

## Remediation

Refer to the linked advisories for vendor-supplied fixes and affected version ranges.

## Package advisory (CVE-2026-114055)

Affected packages:

- `vllm >= 0.7.3, < 0.31.0`

Patched in:

- `vllm 0.31.0`

Source: https://osv.dev/vulnerability/GHSA-gx7p-2j49-hfq4

## Vendor advisories

- **Red Hat VEX** · Important · 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 2026-10-22 · [vex](https://security.access.redhat.com/data/csaf/v2/vex/2026/cve-2026-114055.json)