vLLM from 0.20.0 before 0.30.0 contains a denial of service vulnerability that allows remote attackers to crash the engine by sending inconsistent multimodal metadata to the disaggregated /inference/v1/generate endpoint. Attackers can replay a valid multimodal item with repeated or overlapping mm_placeholders ranges or mismatched kwargs_data and mm_hashes lists, producing a negative text span in Qwen2.5-VL position construction that terminates the engine.
https://github.com/vllm-project/vllm/security/advisories/GHSA-34qw-q2h2-jp3m
Published: 2026-10-25
Updated: 2026-10-25
Base Score: 5.4
Vector: CVSS2#AV:N/AC:H/Au:N/C:N/I:N/A:C
Severity: Medium
Base Score: 5.9
Vector: CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:N/I:N/A:H
Severity: Medium
Base Score: 8.2
Vector: CVSS:4.0/AV:N/AC:H/AT:P/PR:N/UI:N/VC:N/VI:N/VA:H/SC:N/SI:N/SA:N
Severity: High