CVE-2026-22807 — vLLM affected by RCE via auto_map dynamic module loading during model initialization

Description (CNA)

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.

Affected (CNA)

Vendor / product	Versions
vllm-project / vllm	affected >= 0.10.1, < 0.14.0

CWE

CWE-94

References

ADP enrichment

CISA-ADP

No English description supplied.

No references supplied.

redhat-SADP

A flaw was found in vLLM, an inference and serving engine for large language models (LLMs). This vulnerability allows a remote attacker to achieve arbitrary code execution on the vLLM host during model loading. This occurs because vLLM loads Hugging Face `auto_map` dynamic modules without properly validating the `trust_remote_code` setting. By influencing the model repository or path, an attacker can execute malicious Python code at server startup, even before any API requests are handled.