vLLM is an inference and serving engine for large language models (LLMs). Starting in version 0.10.1 and prior to version 0.18.0, two model implementation files hardcode `trust_remote_code=True` when loading sub-components, bypassing the user's explicit `--trust-remote-code=False` security opt-out. This enables remote code execution via malicious model repositories even when the user has explicitly disabled remote code trust. Version 0.18.0 patches the issue.
| Vendor | Product | Versions |
|---|---|---|
| vllm-project | vllm | >= 0.10.1, < 0.18.0 |
Updated description with new details about the Model Handler component and clarified that no exploit is available.
Updated severity to CRITICAL, marked as actively exploited, and specified patch version 0.18.0.
Initial creation
