NetSight
high8.7

CVE-2026-126397

vLLM from 0.10.1 before 0.31.0 forwards a request-controlled mm_processor_kwargs.max_patches value to the Hugging Face processor for Cohere2 Vision models with…

Impact

vLLM from 0.10.1 before 0.31.0 forwards a request-controlled mm_processor_kwargs.max_patches value to the Hugging Face processor for Cohere2 Vision models without a server-side upper bound, causing quadratic CPU work during multimodal preprocessing. Remote clients can send small image chat-completion requests with large, varying max_patches values to bypass the LRU cache, exhaust preprocessing workers, and deny service to other clients.

An attacker can exploit this over the network, without logging in, without any action by a user. If it succeeds: high risk of outages.

Affected products · Vllm-project

ProductAffected versions
Vllm
from 0.10.1 before 0.31.0

Weakness classification

References