CVE-2026-22807
HIGH SEVERITYCVSS Score & Metrics
Base Score
8.8 / 10
Vector String
CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H
Vulnerability Description
vLLM is an inference and serving engine for large language models (LLMs). Starting in version 0.10.1 and prior to version 0.14.0, vLLM loads Hugging Face `auto_map` dynamic modules during model resolution without gating on `trust_remote_code`, allowing attacker-controlled Python code in a model repo/path to execute at server startup. An attacker who can influence the model repo/path (local directory or remote Hugging Face repo) can achieve arbitrary code execution on the vLLM host during model load. This happens before any request handling and does not require API access. Version 0.14.0 fixes the issue.
Vulnerability Details
Published Date
Last Modified
CWE ID
CWE-94
Source
GitHub
Vendor
pip
Product
vllm
External References
- https://github.com/vllm-project/vllm/security/advisories/GHSA-2pc9-4j83-qjmr
- https://github.com/vllm-project/vllm/pull/32194
- https://github.com/vllm-project/vllm/commit/78d13ea9de4b1ce5e4d8a5af9738fea71fb024e5
- https://github.com/vllm-project/vllm/releases/tag/v0.14.0
- https://nvd.nist.gov/vuln/detail/CVE-2026-22807
- https://github.com/advisories/GHSA-2pc9-4j83-qjmr
Discussion (0)
Add Comment
No comments yet. Be the first!