CVE-2025-32444CRITICAL 9.8EPSS p70.3%

CVE-2025-32444CVE-2025-32444

Description

vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs. Versions starting from 0.6.5 and prior to 0.8.5, having vLLM integration with mooncake, are vulnerable to remote code execution due to using pickle based serialization over unsecured ZeroMQ sockets. The vulnerable sockets were set to listen on all network interfaces, increasing the likelihood that an attacker is able to reach the vulnerable ZeroMQ sockets to carry out an attack. vLLM instances that do not make use of the mooncake integration are not vulnerable. This issue has been patched in version 0.8.5.

Scoring

CVSS 3.19.8 (CRITICAL)
VectorCVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H
EPSS1.47% probability of exploitation · percentile 70.3% · 2026-06-19T12:03:05Z
Published2025-04-30
Last modified2025-05-28

Underlying weaknesses· 1

CWE-502

References

  1. https://github.com/vllm-project/vllm/blob/32b14baf8a1f7195ca09484de3008063569b43c5/vllm/distributed/kv_transfer/kv_pipe/mooncake_pipe.py#L179
  2. https://github.com/vllm-project/vllm/commit/a5450f11c95847cf51a17207af9a3ca5ab569b2c
  3. https://github.com/vllm-project/vllm/security/advisories/GHSA-hj4w-hm2g-p6w5
  4. https://github.com/vllm-project/vllm/security/advisories/GHSA-x3m8-f7g5-qhm7

1

TypeTargetConfidenceTier
WeaknessDeserialization of Untrusted Datacwe-5020%live

Related by meaning· 6

Nearest entities by semantic similarity across the cs-graph corpus.

CVE
CVE-2025-29783
CVE
CVE-2025-30165
CVE
CVE-2026-27893
CVE
CVE-2025-62164
CVE
CVE-2025-66448
CVE
CVE-2026-22807
Sourced from NVD + FIRST.org EPSS. Curated for EU compliance use cases by Adam Lundqvist, Founder at SQUR.