CVE-2025-24357
vLLM is a library for LLM inference and serving. vllm/model_executor/weight_utils.py implements hf_model_weights_iterato
vLLM is a library for LLM inference and serving. vllm/model_executor/weight_utils.py implements hf_model_weights_iterator to load the model checkpoint, which is downloaded from huggingface. It uses the torch.load function and the weights_only parameter defaults to False. When torch.load loads malicious pickle data, it will execute arbitrary code during unpickling.
This vulnerability is fixed in v0.7.0.
HIGH · CVSS 7.5
EPSS 0.01009
Schedule remediation
- CVSS base score ≥ 7.0
Sigma rules0
YARA rules0