Description
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.
Affected products
vllm: vllm
References10
- github.com/vllm-project/vllm/security/advisories/GHSA-7972-pg2x… Vendor Advisory
- github.com/vllm-project/vllm/commit/00bd08edeee5dd4d4c13277c011… Patch
- github.com/vllm-project/vllm/pull/36192 Issue Tracking
- access.redhat.com/errata/RHSA-2026:10140
- access.redhat.com/errata/RHSA-2026:10141
- access.redhat.com/errata/RHSA-2026:19712
- access.redhat.com/errata/RHSA-2026:19724
- access.redhat.com/errata/RHSA-2026:19725
- access.redhat.com/errata/RHSA-2026:24977
- access.redhat.com/errata/RHSA-2026:37275
Metrics
- CVSS 3.1
-
8.8
HIGH
Score source: CNA
CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H - EPSS — exploit probability
- 1% percentile: 73 · as of 12 November 2026
- CWE
- CWE-693CWE-501
Could this be exploited in your environment?
A patch closes the known hole. A penetration test shows what an attacker can still reach — with proof.