NetSight
high7.5

CVE-2026-41523

vLLM: Security Check Bypass via assert Statement in Activation Function Loading Allows Arbitrary Code Execution

Impact

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 packages

EcosystemPackageAffectedFixed in
pipvllm< 0.22.00.22.0

Affected products

VendorProductAffected versions
VllmVllm
before 0.22.0

Weakness classification

References