CVE-2025-67729: PYSEC-2026-1578 - lmdeploy vulnerable to Arbitrary Code Execution via Insecure Deserialization in torch.load()

CVE-2025-67729 · Severity: high · CVSS 8.8 · Published 2025-12-26

LMDeploy is a toolkit for compressing, deploying, and serving LLMs. Prior to version 0.11.1, an insecure deserialization vulnerability exists in lmdeploy where torch.load() is called without the weights_only=True parameter when loading model checkpoint files. This allows an attacker to execute arbitrary code on the victim's machine when they load a malicious .bin or .pt model file. This issue has been patched in version 0.11.1.

Technologies: Internlm Lmdeploy, lmdeploy (PyPI). Vendors: Internlm, PyPI.

Quick answers

Is CVE-2025-67729 exploited in the wild?
No. Junglewise Threat Intelligence has no report of CVE-2025-67729 being exploited in the wild.
How severe is CVE-2025-67729?
High severity, with a CVSS score of 8.8, published 26 December 2025.
Which products and versions are affected?
lmdeploy (PyPI) ECOSYSTEM:introduced=0,fixed=0.11.1.
Is there a fix for CVE-2025-67729?
Yes. lmdeploy (PyPI): upgrade to 0.11.1 or later. Junglewise finds every place you run it, drafts the upgrade and applies it once you approve.
Am I affected by CVE-2025-67729?
Check whether you run Internlm Lmdeploy or lmdeploy (PyPI) at an affected version. Junglewise does that check for you around the clock: connect your code, cloud and domains, and it matches new vulnerabilities like this one against what you actually run and tells you when one reaches you.

How to fix

  • lmdeploy (PyPI): upgrade to 0.11.1 or later
    pip install "lmdeploy>=0.11.1"

Affected products

  • Internlm Lmdeploy
  • lmdeploy (PyPI) ECOSYSTEM:introduced=0,fixed=0.11.1

Identifiers

  • CVE-2025-67729
  • GHSA-9pf3-7rrr-x5jh
  • PYSEC-2026-1578

This page as Markdown. Free to reuse under CC BY 4.0 with a link back.