HIGHCVSS v3.1 7.5Vulnerability

CVE-2026-130686

vLLM from 0.6.0 before 0.33.0 lacks a request-wide media budget, allowing remote inference clients to exhaust frontend resources by submitting many remote image URLs in one request. Attackers can send a compact request to /v1/messages or /v1/responses that forces downloading and decoding of every image before prompt-length admission, exhausting memory, CPU, and connections.

Properties

Severity
HIGH
Nvd fetched at
2026-11-19T03:31:17.617109+00:00
Cvss severity
HIGH
Cvss score
7.5
Retrieved at
2026-11-19T03:32:14+00:00
Score
7.5
Cvss version
3.1
Last source
NVD
Cve id
CVE-2026-130686
Cvss vector
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H
First seen by Signal
2026-11-18T15:09:46+00:00
Name
CVE-2026-130686
Vector
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H
Published at
2026-11-18T12:17:12.677
Last modified
2026-11-18T14:17:37.297

Related Entities (1)

HAS_WEAKNESS (1)

Related research

Explore deeper with Ninja Signal's threat intelligence graph

CVE-2026-130686 (CVSS v3.1 7.5) — Ninja Signal Threat Intelligence | Ninja Signal