CVE-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.

Affected Systems

Vendor	vllm
Product	vllm
Affected Versions	from 0.6.5 before 0.8.5
CWE ID	CWE-502

Mitigation

Apply the latest security patches from the vendor, restrict network exposure where applicable, and monitor for exploitation attempts.

Fix Instructions

References

• https://github.com/vllm-project/vllm/blob/32b14baf8a1f7195ca...
• https://github.com/vllm-project/vllm/commit/a5450f11c95847cf...
• https://github.com/vllm-project/vllm/security/advisories/GHS...
• https://github.com/vllm-project/vllm/security/advisories/GHS...