CVE-2026-103041: LightLLM multimodal pickle deserialization in RPyC cache service

CVE-2026-103041 · Severity: critical · CVSS 9.8 · Published 2026-09-29

LightLLM is a Python framework for deploying large language models with multimodal (image and audio) support. In multimodal deployments, LightLLM exposes an unauthenticated remote service that accepts serialized Python objects and deserializes them without validation. An attacker on the network can craft malicious serialized objects to execute arbitrary code on the server with the privileges of the LightLLM service account.

Technologies: ModelTC LightLLM. Vendors: ModelTC.

Quick answers

Is CVE-2026-103041 exploited in the wild?
No. Junglewise Threat Intelligence has no report of CVE-2026-103041 being exploited in the wild.
How severe is CVE-2026-103041?
Critical severity, with a CVSS score of 9.8, published 29 September 2026.
Which products and versions are affected?
ModelTC LightLLM through 1.2.0.
Is there a fix for CVE-2026-103041?
No fixed version is recorded yet. Check the vendor's advisory in the references below. If you run it, Junglewise drafts a mitigation that fits your setup and applies it once you approve, with senior security engineers behind it.
How is CVE-2026-103041 attacked?
It is attacked over the network, in a software library.
Am I affected by CVE-2026-103041?
Check whether you run ModelTC LightLLM 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.

Technical details

The vulnerability exists in the embed-cache RPyC service (lightllm/server/embed_cache/manager.py) which is bound to all interfaces (0.0.0.0) with allow_pickle enabled and no authentication. The exposed_alloc() method passes caller-supplied arguments to obtain(), which deserializes pickled objects on the server. An attacker can exploit Python pickle's __reduce__ protocol to achieve remote code execution by sending crafted objects to any exposed cache method over the network.

Affected products

  • ModelTC LightLLM through 1.2.0

Timeline

  • 2026-09-29: disclosed

References

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