Skip to content
skopnix

NVD published 18 Sep 2026

CVE-2026-33625

LMDeploy is a toolkit for compressing, deploying, and serving large language models. Versions 012.1 through 0.12.2 contain a code injection vulnerability in `lmdeploy/pytorch/config.py` line 620 that allows an attacker to execute arbitrary Python code by publishing a malicious HuggingFace model with a crafted `quantization_config.quant_dtype` value. When a user loads the model with lmdeploy, the `quant_dtype` is passed to `eval(f'torch.{quant_dtype}')` without any validation. Version 0.12.3 contains a patch.

Signals

KEV, exploited in the wild
no
EPSS, 30-day forecast
0.4%, 39th percentile (as of 18 Nov 2026)
CVSS, NVD, HIGH
8.8
Vector
CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H
Internet exposure
not stated

Not in CISA's catalogue as of the last daily pull. Absence is not proof of safety.

Indicators none tied yet

No threat feed has tied an indicator to this CVE yet. Checked every six hours.

On the wire 1 dispatch

  1. GHSA-3hmm-rh5q-gwwr: LMDeploy vulnerable to arbitrary code execution via eval() of untrusted quant_dtype in model config loading

Public exploit code none in our sources

No public exploit in the sources we track: GitHub proof-of-concept lists, Metasploit and Exploit-DB. Absence is not proof that none exists.

References

Watch this one?

KEV additions reach Telegram and the KEV-only RSS as they land.

Telegram ↗KEV RSSAPIStack check