CVE-2026-44222

Description

vLLM is an inference and serving engine for large language models (LLMs). From 0.6.1 to before 0.20.0, there is a a Token Injection vulnerability in vLLM’s multimodal processing. Unauthenticated, text-only prompts that spell special tokens are interpreted as control. Image and video placeholder sequences supplied without matching data cause vLLM to index into empty grids during input-position computation, raising an unhandled IndexError and terminating the worker or degrading availability. Multimodal paths that rely on image_grid_thw/video_grid_thw are affected. This vulnerability is fixed in 0.20.0.

Severity
MEDIUM
CVSS v3
6.5
CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H · source cna
Exploit chance (EPSS)
0.414%
chance of exploitation in the next 30 days
higher than 36.3% of scored CVEs · 2026-11-17
CISA KEV
Not listed
Public exploit code
Yes
Exploit references (NVD and CNA): 1
Metasploit modules: 0
Nuclei templates: 0

Weakness

CISA SSVC decision points

Exploitationnone
Automatableno
Technical impactpartial
Scored2026-05-13

Timeline

  1. 2026-05-05GitHub advisory published (GHSA-hpv8-x276-m59f)7 days before publication
  2. 2026-05-12Published by the CNA19:57 UTC
  3. 2026-05-13CISA SSVC decision recorded1 day after publication

The timeline shows what already happened. To be told when this CVE changes next, watch it from the API.

Exploit and advisory records

GitHub advisory GHSA-hpv8-x276-m59f, affected packages:

PackageVulnerable versionsFixed in
PyPI vllmfrom 0.6.1 before 0.20.00.20.0

Affected products

Products listed as vulnerable. Upgrade to the version under "Fixed in", or a later one.

ProductVulnerable versionsFixed in
vllm vllmfrom 0.6.1 before 0.20.0fixed in 0.20.00.20.0

Products and versions as NVD and the vendor that published the CVE list them.

References

Get more CVE data from the API.

Free without a key or signup. Pro adds monitoring and higher limits: pricing.

Sources