CVE-2026-41523

Medium

Elevated severity or exploit probability.

CVSS base

7.5 HIGH

CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:U/C:H/I:H/A:H

EPSS — probability of exploitation (30 days)

0.9%

58.4th percentile

CISA KEV

Not listed

Weakness / dates

CWE-617

Published 2026-06-22 · modified 2026-09-04

CVSS breakdown

CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:U/C:H/I:H/A:H

Attack Vector	N	Network
Attack Complexity	H	High
Privileges Required	N	None
User Interaction	R	Required
Scope	U	Unchanged
Confidentiality	H	High
Integrity	H	High
Availability	H	High

Timeline

2026-06-22 — Published (NVD)
2026-09-04 — Last modified (NVD)

Description

vLLM is an inference and serving engine for large language models (LLMs). Prior to 0.22.0, an assert-based security check in vLLM's activation function loading allows any unauthenticated attacker to achieve arbitrary code execution on the server by publishing a malicious HuggingFace model, when vLLM runs in Python optimized mode (python -O or PYTHONOPTIMIZE=1). This vulnerability is fixed in 0.22.0.

Affected

vllm

References

Official: NVD · CVE.org