Severity by source
AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H
Unauthenticated network-reachable RPC deserializing attacker pickle gives direct RCE with full host compromise; randomized port is a discovery hurdle, not CVSS access complexity, so AC:L.
Primary rating from Vendor (GitHub_M).
CVSS VectorVendor: GitHub_M
Lifecycle Timeline
5DescriptionCVE.org
LMDeploy is a toolkit for compressing, deploying, and serving large language models. Starting in version 0.9.1 and prior to version 0.10.2, the LMdeploy implements an rpc server (AsyncRPCServer in zmq_rpc.py) for supporting the RPC communications. In its core functionality call_and_response(), I found it will directly use the pickles.loads() to deserialize the received messages without any sanitization, hence resulting in a remote code execution vulnerability by this RPC server. Version 0.10.2 contains a patch.
Articles & Coverage 1
AnalysisAI
Remote code execution in InternLM LMDeploy versions 0.9.1 through 0.10.1 allows unauthenticated network attackers to run arbitrary commands by sending crafted pickle payloads to the AsyncRPCServer RPC endpoint (zmq_rpc.py). The service deserializes raw messages with pickle.loads() and performs no authentication or IP validation, so any attacker who can reach the randomized port-typically found by scanning-can trigger code execution; publicly available exploit code exists (proof-of-concept with reverse shell demo), and the CVSS 3.1 score is 9.8. This is not confirmed actively exploited (no CISA KEV listing), and exploitation is not possible if the RPC service is bound exclusively to localhost; upgrading to 0.10.2 patches the deserialization flaw and restricts the RPC to localhost.
Technical ContextAI
LMDeploy uses a ZeroMQ-based RPC layer for communication between its components; the AsyncRPCServer class in zmq_rpc.py handles incoming requests in call_and_response(), which calls pickle.loads() directly on the received bytes without any sanitization or class whitelist. This is a textbook CWE-502 deserialization-of-untrusted-data flaw: pickle is not a safe format for untrusted input, and its __reduce__ mechanism allows an attacker to instantiate arbitrary classes and execute commands during deserialization. The affected package is pip/lmdeploy, identified by CPE cpe:2.3:a:internlm:lmdeploy:*:*:*:*:*:*:*:* (the CPE does not encode version granularity, but the vendor confirms the vulnerable range is >=0.9.1, <0.10.2). The RPC server lacks authentication and does not validate the source IP, and although the listening port is randomized, an attacker on the network can port-scan to locate it; version 0.10.2 contains the vendor patch, with the release notes specifically citing a change to bind zmqrpc to localhost only (PR #4017).
RemediationAI
Upgrade to LMDeploy 0.10.2 (vendor-released patch) as the primary fix; this version patches the deserialization exposure and binds the zmqrpc server to localhost only (PR #4017), which prevents remote exploitation. If immediate upgrade is not possible, restrict the AsyncRPCServer to localhost or a trusted network segment by configuring the RPC binding and/or blocking its dynamically assigned port with host or network firewall rules-note that the randomized port requires scanning to find, so closing all unnecessary ports reduces exposure. Alternatively, disable the RPC service entirely if it is not required for your deployment; trade-off is loss of distributed inference functionality. Monitor for unexpected port scans or connections to the LMDeploy RPC port. Refer to the vendor advisory (https://github.com/InternLM/lmdeploy/security/advisories/GHSA-5h8j-6crg-7rmw) and release notes (https://github.com/InternLM/lmdeploy/releases/tag/v0.10.2) for details.
Wazuh SIEM platform versions 4.4.0 through 4.9.0 contain an unsafe deserialization vulnerability in the DistributedAPI t
BentoML version 1.4.2 and earlier contains an unauthenticated remote code execution vulnerability through insecure deser
pgAdmin 4 contains critical remote code execution vulnerabilities in the Query Tool download and Cloud Deployment endpoi
The renderLocalView function in render/views.py in graphite-web in Graphite 0.9.5 through 0.9.10 uses the pickle Python
BentoML is a Python library for building online serving systems optimized for AI apps and model inference. Rated critica
OpenSSL before 0.9.8za, 1.0.0 before 1.0.0m, and 1.0.1 before 1.0.1h does not properly restrict processing of ChangeCiph
pyLoad download manager version prior to 0.5.0b3.dev77 exposes the Flask SECRET_KEY through an unauthenticated endpoint.
Langflow (a visual LLM pipeline builder) contains a critical unauthenticated code execution vulnerability (CVE-2026-3301
In Mercurial before 4.1.3, "hg serve --stdio" allows remote authenticated users to launch the Python debugger, and conse
Unauthenticated remote code execution affects Kestra OSS (the open-source event-driven orchestration platform) prior to
Unauthenticated remote code execution in Marimo ≤0.20.4 allows attackers to execute arbitrary system commands via the `/
pyLoad is the free and open-source Download Manager written in pure Python. Rated medium severity (CVSS 5.3), this vulne
Same weakness CWE-502 – Deserialization of Untrusted Data
View allShare
External POC / Exploit Code
Leaving vuln.today
EUVD-2025-210921
GHSA-5h8j-6crg-7rmw