CVE-2025-32444: vLLM Vulnerable to Remote Code Execution via Mooncake Integration
Impacted Deployments
Note that vLLM instances that do NOT make use of the mooncake integration are NOT vulnerable.
Description
vLLM integration with mooncake is vaulnerable 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.
This is a similar to GHSA - x3m8 - f7g5 - qhm7, the problem is in
https://github.com/vllm-project/vllm/blob/32b14baf8a1f7195ca09484de3008063569b43c5/vllm/distributed/kvtransfer/kvpipe/mooncakepipe.py#L179
Here recvpyobj() Contains implicit pickle.loads(), which leads to potential RCE.
Other sources
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.
— MITRE
Affected Software
Remediation
Event History
Frequently Asked Questions
What is the severity of CVE-2025-32444?
CVE-2025-32444 is classified as a critical vulnerability due to its potential for remote code execution.
How do I fix CVE-2025-32444?
To mitigate CVE-2025-32444, upgrade to a version of vLLM that is greater than 0.8.5 and ensure that mooncake integration is not in use.
Which versions of vLLM are affected by CVE-2025-32444?
Versions of vLLM from 0.6.5 to 0.8.5 are vulnerable to CVE-2025-32444.
What types of attacks can CVE-2025-32444 facilitate?
CVE-2025-32444 can facilitate remote code execution attacks due to the insecure use of `pickle` serialization over ZeroMQ.
What systems are primarily affected by CVE-2025-32444?
Only vLLM instances that utilize the mooncake integration are affected by CVE-2025-32444, while others remain secure.