Skip to main content
Public proof of conceptFix available

CVE-2026-22778

A public proof of concept exists; no exploitation attested.

  • Published2026-02-02
  • Fixed in0.14.1
  • WeaknessCWE-209 · CWE-532

Am I affected?

With a domain, we read the version its public headers and pages expose. A distribution build can carry the fix under an older number.

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.

Timeline

dates in UTC
  1. 2026-02-0223:16CVE publishedNVD

Versions per branch

NVD · OSV
Vllm
BranchAffectedFixed in
0.x≥ 0.8.3, < 0.14.10.14.1

as publishedRanges as NVD and OSV publish them; a distribution’s build can be fixed under an older number.

Products declared by the CNA1 product

As GitHub_M, the CNA that assigned this CVE, declared them in the CVE record; versions as written.

  • vllm-project vllm
    Affected
    • = >= 0.8.3, < 0.14.1no fix stated

Source: the CNA record (CVE Program), © The MITRE Corporation,CVE Terms of UseCVE recordupdated 2026-02-02

Distribution builds and vendor advisories

Packages from OSV.dev and products from the vendors' CSAF advisories.

version not verifiedVersions not verified: read as OSV.dev and the vendors publish them, not checked against each vendor's release list.

Packages

  • tritonserver-backend-vllm-cuda-12.9Chainguardall versions · fixed in 25.9.0_git20260318-r0
  • tritonserver-backend-vllm-cuda-12.9Chainguardall versions · fixed in 25.9.0_git20251112-r7
  • tritonserver-backend-vllm-cuda-12.9Chainguardall versions · fixed in 25.9.0_git20251112-r6
  • tritonserver-backend-vllm-cuda-13.0Chainguardall versions · fixed in 25.11-r2

Vendor advisories

  • CVE-2026-22778 Red Hat Product Security18 products fixed · 4 known affected

    vLLM: vLLM: Remote code execution via invalid image processing in the multimodal endpoint.

    rhelai3/bootc-aws-cuda-rhel9 as a component of Red Hat Enterprise Linux AI (RHEL AI) 3 · rhelai3/bootc-azure-cuda-rhel9 as a component of Red Hat Enterprise Linux AI (RHEL AI) 3 · rhelai3/bootc-cuda-rhel9 as a component of Red Hat Enterprise Linux AI (RHEL AI) 3 · +1 more

  • RHSA-2026:30089 Red Hat Product Security2 products fixed

    Red Hat Security Advisory: Red Hat AI Inference Server 3.3.5 (CUDA)

  • RHSA-2026:30088 Red Hat Product Security1 product fixed

    Red Hat Security Advisory: Red Hat AI Inference Server 3.3.5 (ROCm)

  • RHSA-2026:30087 Red Hat Product Security3 products fixed

    Red Hat Security Advisory: Red Hat AI Inference Server 3.3.5 (Spyre)

  • RHSA-2026:19712 Red Hat Product Security2 products fixed

    Red Hat Security Advisory: RHOAI 3.3.3 - Red Hat OpenShift AI

  • RHSA-2026:3782 Red Hat Product Security6 products fixed

    Red Hat Security Advisory: RHOAI 2.25.3 - Red Hat OpenShift AI

3 more advisories

Full record on NVD

Fix

  1. Update Vllm to 0.14.1
  2. Vendor fix Red Hat Product Security RHSA-2026:30089 · 2 productsFor more information visit https://access.redhat.com/errata/RHSA-2026:30089
  3. Vendor fix Red Hat Product Security RHSA-2026:30088 · 1 productFor more information visit https://access.redhat.com/errata/RHSA-2026:30088
  4. Vendor fix Red Hat Product Security RHSA-2026:30087 · 3 productsFor more information visit https://access.redhat.com/errata/RHSA-2026:30087
  5. Vendor fix Red Hat Product Security RHSA-2026:19712 · 2 productsFor Red Hat OpenShift AI 3.3.3 see the following documentation, which will be updated shortly for this release, for important instructions on how to upgrade your cluster and fully apply this errata update: https://docs.redhat.com/en/documentation/red_hat_openshift_ai/
  6. Vendor fix Red Hat Product Security RHSA-2026:3782 · 6 productsFor Red Hat OpenShift AI 2.25.3 see the following documentation, which will be updated shortly for this release, for important instructions on how to upgrade your cluster and fully apply this errata update: https://docs.redhat.com/en/documentation/red_hat_openshift_ai/
  7. Vendor fix Red Hat Product Security RHSA-2026:3713 · 1 productFor Red Hat OpenShift AI 3.3 see the following documentation, which will be updated shortly for this release, for important instructions on how to upgrade your cluster and fully apply this errata update: https://docs.redhat.com/en/documentation/red_hat_openshift_ai/

Advisories and patches

Detection

read 2026-11-01 01:02 UTC
Nuclei template
available
ProjectDiscovery
Exploit-DB
no entry
Exploit-DB
Proof of concept
Metasploit modules
1 moduleauxiliary/scanner/http/vllm_multimodal_info_leakrank: normalModule metadata: Metasploit Framework, © Rapid7, Inc., BSD-3-Clause.
Metasploit Framework

Sources

Identifiers

Technical details

CVSS v3 Vector
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H
Published
2026-02-02
Last Modified
2026-07-15
NVD status
Modified
GitHub Advisory
GHSA-4r2x-xpjr-7cvv

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 18.3% on 2026-11-08, 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), with vector string CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H.

Running vllm? Hear about its next vulnerability

Surveillance reads the versions in your stack — a domain, a lockfile or a list — checks each CVE against the ranges its sources publish, and alerts you when one you run is exploited or enters CISA KEV. A version it cannot read stays « version not verified ».

Get alerts for this productCISA KEV additions, November 2026Search CVEs

No credit card required · No monthly cap on website lookups · 250 API requests/month

Go further

Related vulnerabilities

Same product first, then the same vendor advisory and the same publication week.

Back to CVE Database

Sources for this record