CVE-2026-56340: vllm from 0.10.2 (inclusive), before 0.13.0 (exclusive); +4 more affected products

Affected technology

vllm · from 0.10.2 (inclusive), before 0.13.0 (exclusive)
vllm

vLLM · 0.10.2 to before 0.13.0
vLLM

Red Hat AI Inference Server · Exact affected versions not specified by the source
Red Hat

Red Hat Enterprise Linux AI (RHEL AI) 3 · Exact affected versions not specified by the source
Red Hat

Red Hat OpenShift AI (RHOAI) · Exact affected versions not specified by the source
Red Hat

Description’s affected range: versions >= 0.10.2 and < 0.13.0 are missing sparse tensor validation in multimodal embeddings processing

Component: Not specified by the source

Attack conditions (Vendor/CNA, CVSS 4.0): Network (remote) · Low privileges required · No user interaction required

What an attacker can do

An attacker can submit crafted embedding requests with malformed (negative or out-of-bounds) tensor indices, when the prompt-embeds feature is enabled, to trigger crashes or resource exhaustion (denial of service), with potential for out-of-bounds/write-what-where memory corruption. Vendor/CNA’s CVSS 4.0 assessment (base score 8.7/10) rates confidentiality, integrity and availability impact as high.