LMDeploy has Remote Code Execution by Pickle Deserialization via handle_zmq_recv in lmdeploy/lmdeploy/pytorch/disagg/conn/engine_conn.py (CVE-2025-66455) | HOL Guard CVE

Critical · CVSS 9.8CVE-2025-66455GHSA-2VH9-42VM-XMV2

LMDeploy has Remote Code Execution by Pickle Deserialization via handle_zmq_recv in lmdeploy/lmdeploy/pytorch/disagg/conn/engine_conn.py

Answer in brief

CVE-2025-66455 records a Critical severity (CVSS 9.8) vulnerability in LMDeploy has Remote Code Execution by Pickle Deserialization via handle_zmq_recv in lmdeploy/lmdeploy/pytorch/disagg/conn/engine_conn.py. The current sources do not mark it as known exploited. The current feed maps InternLM/lmdeploy (generic), lmdeploy (pip). Check affected ranges and fixed versions before updating.

Analysis pending evidence review

HOL Guard separates source facts from reviewed analysis. See the methodology.

Published Sep 18, 2026Updated Sep 18, 2026Source checked Sep 18, 2026First seen by HOL Sep 18, 2026Material review Sep 18, 2026

Upstream Advisory

Key facts

Risk: Critical · CVSS 9.8
Exploitation: Not marked as known exploited
Affected software: 2 mapped packages or products
Fix availability: Available

Why this deserves its current priority

CVSS is 9.8. The current sources do not mark it as known exploited. Treat this as a source-backed prioritization signal, not a statement about your environment.

Analysis status

Analysis pending evidence review

Factual feed record only; HOL analysis is not approved for indexing. Read the methodology.

Affected scope and exposure questions

The current feed maps InternLM/lmdeploy (generic), lmdeploy (pip). Check affected ranges and fixed versions before updating.

Mapped affected packages and fixed versions
Package	Affected range	Fixed version
InternLM/lmdeploygeneric	>= 0.9.2, < 0.16.0	Not reported
lmdeploypip	>=0.9.2,<0.16.0	0.16.0

Recommended response

Check inventory. Check lockfiles and deployed manifests for InternLM/lmdeploy, lmdeploy.
Review the reported fix. Update lmdeploy to 0.16.0 if you use the affected versions. Test the change in a non-production environment first.

Evidence timeline and material changes

Published upstream
Sep 18, 2026
Evidence: source:cvelist:source_dates:source-dates:record
Source modified
Sep 18, 2026
Evidence: source:cvelist:source_dates:source-dates:record
First seen by HOL
Sep 18, 2026

Sources and claim methodology

Upstream source description

LMDeploy is a toolkit for compressing, deploying, and serving large language models. Starting in version 0.9.2 and prior to version 0.16.0, LMDeploy's PyTorch DistServe/PD-disaggregation control plane used `recv_pyobj()` to deserialize messages received through a ZeroMQ PULL socket. PyZMQ implements `recv_pyobj()` using Python pickle deserialization, which can execute arbitrary code while reconstructing an object. The peer address used by the receiver was supplied through the `POST /distserve/p2p_connect` HTTP endpoint. An attacker who could reach an affected DistServe API server could cause the server to connect to an attacker-controlled ZeroMQ endpoint and deserialize a crafted pickle payload. API-key authentication is not enabled unless the operator explicitly configures it. As a result, affected DistServe deployments without API keys allowed unauthenticated remote code execution with the privileges of the LMDeploy serving process. This issue affects the PyTorch backend when PD-disaggregation/DistServe is enabled. Ordinary deployments that do not use the affected disaggregated-serving path do not expose this data flow. The fix was released in LMDeploy 0.16.0. Users who cannot upgrade immediately should prevent untrusted clients from reaching `/distserve/*` endpoints, restrict the DistServe HTTP and ZeroMQ control planes to trusted cluster networks, configure API-key authentication, and block arbitrary outbound ZeroMQ connections from serving nodes. These measures reduce exposure but do not make pickle deserialization safe.

Quoted source text, attributed separately from HOL analysis.