vLLM: Prior to 0.30.0, the Rust frontend's track_http_metrics middleware records the raw HTTP method…
vLLM
More for vLLM RSS feed for vLLM
- Severity
- High, CVSS 7.5
- CVSS vector
- CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H
- Record
- CVE-2026-105759
- Affects
- vLLM
- Known exploited
- Not listed in the CISA Known Exploited Vulnerabilities catalog
- Published
- Record first published by the source
- October 5, 2026
- Source last modified
- October 16, 2026
Description
vLLM is an inference and serving engine for large language models. Prior to 0.30.0, the Rust frontend's track_http_metrics middleware records the raw HTTP method token as a Prometheus label for requests reaching registered routes. An unauthenticated attacker can send unique arbitrary method tokens to unguarded routes such as /tokenize, causing Prometheus's Family::get_or_create function to permanently create counter and histogram label sets. Those label sets increase process memory usage and enlarge the /metrics response until the service or monitoring path is exhausted. This issue is fixed in version 0.30.0.
Description quoted from the NVD record.
Who is affected
vLLM is software that runs AI language models on a server so other apps and people can send them questions. This flaw affects versions before 0.30.0, in the part of vLLM that receives incoming web requests and keeps usage statistics for monitoring. Someone without any login can send many web requests with made-up request types, and each one makes vLLM permanently keep a new set of statistics. This steadily uses up memory and makes the monitoring statistics page grow until the service or its monitoring stops working. This flaw is rated High.
What to do
- Find out who runs vLLM for your organization: the person who installed it, or the IT provider or vendor who manages it for you.
- If someone in your organization runs it directly, have them update vLLM to version
0.30.0or later, which fixes this flaw. - If an IT provider or vendor runs vLLM for you, send them this advisory and ask when the update to
0.30.0will be done. - If you are not sure which version you have, ask the person who runs vLLM to check it, since every version before
0.30.0is affected.
Analysis written from the source record as of October 16, 2026 and reviewed before publishing.