vllm: Malicious model to RCE by torch.load in hf_model_weights_iterator

Exploitation

No confirmed exploitation recorded. Patch Next marks exploitation only from CISA KEV.

CVSS

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

EPSS

• CVE-2025-24357: 0.89% estimated exploitation activity in the next 30 days; percentile 59.51 (FIRST EPSS, 2026-11-03). This is a model estimate, not proof of exploitation or customer exposure.

As published by the source; long descriptions are shortened.

Description The vllm/modelexecutor/weightutils.py implements hfmodelweightsiterator to load the model checkpoint, which is downloaded from huggingface. It use torch.load function and weightsonly parameter is default value False. There is a security warning on https://pytorch.org/docs/stable/generated/torch.load.html, when torch.load load a malicious pickle data it will execute arbitrary code during unpickling. Impact This vulnerability can be exploited to execute arbitrary codes and OS commands in the victim machine who fetch the pretrained repo remotely. Note that most models now use the safetensors format, which is not vulnerable to this issue. References https://pytorch.org/docs/stable/generated/torch.load.html Fix: https://github.com/vllm-project/vllm/pull/12366

Published by the source

27 Jan 2025, 20:50 UTC

Last modified by the source

7 Aug 2026, 08:12 UTC

Source last checked

26 Oct 2026, 13:32 UTC

1. Revision 2 · amended · 26 Oct 2026

   Amended from ghsa record GHSA-rh4j-5rhw-hr54

2. Revision 1 · created · 24 Oct 2026

   Published from osv record GHSA-rh4j-5rhw-hr54

Preview the exact text before you copy it. Commands are read-only suggestions; review them before running anything.

# Patch Next advisory brief: CVE-2025-24357

Advisory: vllm: Malicious model to RCE by torch.load in hf_model_weights_iterator
IDs: CVE-2025-24357, GHSA-rh4j-5rhw-hr54
Record: https://patchnext.com/advisories/cve-2025-24357 (revision 2, source checked 26 Oct 2026, 13:32 UTC)
Inventory: none attached. This brief comes from a public page and does not say whether any deployment is affected.
Severity: High (CVSS 7.5)
Exploitation: no confirmed exploitation recorded. Patch Next confirms exploitation only from CISA KEV.
Vendor response: Vendor fix available; fixed versions by package: vllm 0.7.0. A vendor fix does not mean a deployment is fixed.

## Affected components
- PyPI vllm: affected < 0.7.0; fixed in 0.7.0

## Suggested inspection (read-only)
- `python -m pip show vllm`: prints the installed version. Run with the environment that serves production active.

## Supported action
Upgrade each affected component to a fixed version on its release line. For a transitive dependency, upgrade the parent package or add an override or constraint, then regenerate the lockfile. Redeploy every environment that ran an affected version.

## Required regression checks
- Run the test suite and a production build.
- Exercise the features that depend on vllm.
- Repeat the inspection and confirm that only fixed versions remain installed.

## Rollback considerations
- Keep the previous lockfile and deployment artifact. Rolling back reinstates the vulnerable version, so apply any documented mitigation while you fix forward.

## Unresolved questions
- Which deployments run an affected version?
- Exploitation status can change. Check again before lowering the priority.
- Mark the work resolved only after the fixed version is verified in production.

## Sources
- Verified: https://osv.dev/vulnerability/GHSA-rh4j-5rhw-hr54 (OSV record GHSA-rh4j-5rhw-hr54 lists affected ranges: PyPI vllm >=0 <0.7.0 (fixed))
- Verified: https://github.com/advisories/GHSA-rh4j-5rhw-hr54 (GitHub-reviewed advisory GHSA-rh4j-5rhw-hr54 lists affected ranges: PyPI vllm >=0 <0.7.0)

Commands are suggestions: review each one before running it.