CVE-2026-94626: vLLM Memory Exhaustion Vulnerability Due to tp_size Parameter Validation Failure
Summary
CVE-2026-94626: vLLM through 0.29.0 fails to validate the tp_size parameter in kv_transfer_params on OpenAI-compatible completion endpoints, allowing attackers to allocate unbounded memory. Affects vllm versions < 0.30.0. Fixed in 0.30.0. How to fix ↓
Check if this vulnerability affects your application
Upload a manifest or SBOM and we'll show whether you're affected, plus every other risky package in the same scan.
No signup required · Results in about 60 seconds
Background
vLLM is an open-source library designed for efficient memory management and processing of large language models, particularly in OpenAI-compatible environments. It allows users to perform tasks such as text completion and decoding with optimized resource usage. However, the library has a critical vulnerability that arises from improper validation of the tp_size parameter, which can lead to severe memory exhaustion issues.
Vulnerability Detail
The vulnerability in vLLM, identified as CVE-2026-94626, stems from the failure to validate the tp_size parameter in the kv_transfer_params function on OpenAI-compatible completion endpoints. This oversight allows attackers to supply arbitrary tp_size values, leading to unbounded memory allocation. In scenarios where prefill or decode disaggregated deployments are utilized, this can result in memory exhaustion, ultimately triggering the kernel's Out-Of-Memory (OOM) killer to terminate the decode worker process. The implications of this vulnerability are significant, as it can disrupt service availability and degrade performance.
How To Fix
Latest published version: 0.32.0 as of 22 Oct 2026
pip install --upgrade "vllm==0.32.0"Upgrading to it may include breaking changes; the version below is the minimum fixed release for this vulnerability.
If you need the minimum change
0.30.0
This is the first release that contains the fix.
pip install --upgrade "vllm==0.30.0"Tip — Not sure which version your app runs? Vulert can match this fix to it. Scan free →
Workaround
As a temporary workaround, users can implement input validation on the tp_size parameter before passing it to the kv_transfer_params function. This can help mitigate the risk of memory exhaustion until the package is updated. For example, you can add a check like this:
if tp_size < 0 or tp_size > MAX_ALLOWED_SIZE: raise ValueError('Invalid tp_size value')
Conclusion
The vLLM library has a critical vulnerability (CVE-2026-94626) that can lead to memory exhaustion due to improper validation of the tp_size parameter. Users are strongly advised to update to version 0.30.0 or later to mitigate this risk. Implementing input validation can serve as a temporary workaround until the update is applied.
See every vulnerability recorded for vllm →
FAQ
CVE-2026-94626 is a vulnerability in the vLLM library that allows attackers to exploit the tp_size parameter, leading to unbounded memory allocation.
You can use the Vulert playground at Vulert Scanner to check if your application is affected by this vulnerability without needing to sign up.
The vulnerability can lead to memory exhaustion, causing the kernel to terminate the decode worker process, which may disrupt service availability.
You can update the vLLM package by running the command pip install --upgrade vllm in your terminal.
Yes, you can implement input validation on the tp_size parameter to prevent invalid values from being processed.
Related vulnerabilities
Recent vulnerabilities recorded for vllm.
