EUVD-2025-200115

| CVE-2025-66448 HIGH
7.1
CVSS 3.1
Share

CVSS Vector

CVSS:3.1/AV:N/AC:H/PR:L/UI:R/S:U/C:H/I:H/A:H
Attack Vector
Network
Attack Complexity
High
Privileges Required
Low
User Interaction
Required
Scope
Unchanged
Confidentiality
High
Integrity
High
Availability
High

Lifecycle Timeline

4
EUVD ID Assigned
Mar 15, 2026 - 13:34 euvd
EUVD-2025-200115
Analysis Generated
Mar 15, 2026 - 13:34 vuln.today
Patch Released
Mar 15, 2026 - 13:34 nvd
Patch available
CVE Published
Dec 01, 2025 - 23:15 nvd
HIGH 7.1

Description

vLLM is an inference and serving engine for large language models (LLMs). Prior to 0.11.1, vllm has a critical remote code execution vector in a config class named Nemotron_Nano_VL_Config. When vllm loads a model config that contains an auto_map entry, the config class resolves that mapping with get_class_from_dynamic_module(...) and immediately instantiates the returned class. This fetches and executes Python from the remote repository referenced in the auto_map string. Crucially, this happens even when the caller explicitly sets trust_remote_code=False in vllm.transformers_utils.config.get_config. In practice, an attacker can publish a benign-looking frontend repo whose config.json points via auto_map to a separate malicious backend repo; loading the frontend will silently run the backend’s code on the victim host. This vulnerability is fixed in 0.11.1.

Analysis

vLLM is an inference and serving engine for large language models (LLMs). Prior to 0.11.1, vllm has a critical remote code execution vector in a config class named Nemotron_Nano_VL_Config. When vllm loads a model config that contains an auto_map entry, the config class resolves that mapping with get_class_from_dynamic_module(...) and immediately instantiates the returned class. This fetches and executes Python from the remote repository referenced in the auto_map string. Crucially, this happens even when the caller explicitly sets trust_remote_code=False in vllm.transformers_utils.config.get_config. In practice, an attacker can publish a benign-looking frontend repo whose config.json points via auto_map to a separate malicious backend repo; loading the frontend will silently run the backend’s code on the victim host. This vulnerability is fixed in 0.11.1.

Technical Context

Remote code execution allows an attacker to run arbitrary commands or code on the target system over a network without prior authentication. This vulnerability is classified as Code Injection (CWE-94).

Affected Products

Affected products: Vllm Vllm

Remediation

A vendor patch is available — apply it immediately. Apply vendor patches immediately. Restrict network access to vulnerable services. Implement network segmentation and monitoring for anomalous activity.

Priority Score

36
Low Medium High Critical
KEV: 0
EPSS: +0.2
CVSS: +36
POC: 0

Vendor Status

Debian

Bug #1095237
vllm
Release Status Fixed Version Urgency
open - -

Share

EUVD-2025-200115 vulnerability details – vuln.today

This site uses cookies essential for authentication and security. No tracking or analytics cookies are used. Privacy Policy