HIGH · 8.8

CVE-2026-22807

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 resolu...

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.

CVSS Score

8.8

HIGH

CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H
Attack Vector
NETWORK
Attack Complexity
LOW
Privileges Required
NONE
User Interaction
REQUIRED
Scope
UNCHANGED
Confidentiality
HIGH
Integrity
HIGH
Availability
HIGH

Affected Products

VendorProductVersions
VllmVllm>= 0.10.1, < 0.14.0

Related Weaknesses (CWE)

References

FAQ

What is CVE-2026-22807?

CVE-2026-22807 is a vulnerability with a CVSS score of 8.8 (HIGH). 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 resolu...

How severe is CVE-2026-22807?

CVE-2026-22807 has been rated HIGH with a CVSS base score of 8.8/10. Review the CVSS metrics above for detailed severity breakdown.

Is there a patch for CVE-2026-22807?

Check the references section above for vendor advisories and patch information. Affected products include: Vllm Vllm.