vLLM Vulnerabilities

pipLLM Inference

AI Threat Alert tracks 188 known vulnerabilities in vLLM, 13 rated critical — an AI/ML llm inference in the pip ecosystem. Each CVE includes CVSS severity, EPSS exploit probability, patch status, and CISO-grade analysis.

61
Risk Score
188
Total CVEs
13
Critical
pip
Ecosystem
Nov 18, 2026
Last CVE
26%
Patch Rate
37d
Avg Time to Patch
96,942 stars24,834 forks10,395 issues93 dependentsLast push Nov 15, 2026
View on GitHub

Known Vulnerabilities(188 total, page 4 of 8)

SeverityCVE IDSummaryCVSSPublished
MEDIUMCVE-2026-93989vLLM: bad_words OOB index corrupts cross-request logits4.3Sep 19, 2026MEDIUMCVE-2026-93841vLLM: OOB write in penalty kernel corrupts requests5.3Sep 18, 2026MEDIUMCVE-2026-93840vLLM: token ID validation flaw corrupts cross-request logits5.3Sep 18, 2026HIGHCVE-2026-93592vLLM: negative token ID crashes embeddings engine7.5Sep 18, 2026HIGHCVE-2026-93436vLLM: memory leak in PD disaggregation causes DoS7.5Sep 17, 2026HIGHCVE-2026-92925Redis: cluster bus OOB read enables DoS and info leak7.1Sep 17, 2026MEDIUMCVE-2026-69147vLLM: request-selected video backend exhausts GPU memory6.5Sep 16, 2026MEDIUMCVE-2026-57173vLLM: unbounded audio decode causes OOM DoS6.5Sep 16, 2026MEDIUMCVE-2026-92365vLLM: algorithmic complexity DoS in thinking budget4.3Sep 16, 2026MEDIUMCVE-2026-92220vLLM: MoRIIO KV handler flaw allows remote DoS5.3Sep 16, 2026MEDIUMCVE-2026-90878vLLM: chat_template abuse causes resource exhaustion4.3Sep 15, 2026LOWCVE-2026-90713vLLM: local DoS via malformed tiktoken vocab file3.3Sep 14, 2026MEDIUMCVE-2026-90555vLLM: FLAC header spoofing crashes transcription API6.5Sep 12, 2026HIGHCVE-2026-90554vLLM: audio decode DoS via NanoNemotronVL video input7.5Sep 12, 2026HIGHCVE-2026-90553vLLM: RCE bypasses trust_remote_code model safeguard7.8Sep 12, 2026HIGHGHSA-wfgq-w7cq-qj7jmistral.rs: SSRF & file-read via image/audio URLs7.2Sep 10, 2026HIGHCVE-2026-37237vLLM: unbounded media URLs cause memory-exhaustion DoS7.5Aug 28, 2026MEDIUMCVE-2026-18393FFmpeg: heap overflow via crafted TDSC cursor in video5.4Aug 28, 2026MEDIUMCVE-2026-78684vLLM: unauthenticated DoS via DeepStream decode bypass5.3Aug 25, 2026HIGHGHSA-j659-8xh6-5pq5atomic-agents: unpriced models silently bypass cost cap--Aug 17, 2026MEDIUMCVE-2026-73560vLLM: SSRF/LFI bypass in MiMo-V2 multimodal input6.5Aug 17, 2026MEDIUMCVE-2026-71486vLLM: unbounded /derender inputs cause resource DoS4.3Aug 17, 2026MEDIUMCVE-2026-73559vLLM: unbounded prompt batch triggers single-request DoS6.5Aug 13, 2026MEDIUMCVE-2026-73558vLLM: integer overflow leaks other users' output5.3Aug 13, 2026MEDIUMCVE-2026-73557vLLM: race condition bypasses prompt-embeds tensor guard--Aug 13, 2026

Showing 76–100 of 188

Frequently asked questions

What is vLLM?

vLLM is an AI/ML llm inference tracked by AI Threat Alert for security vulnerabilities in the pip ecosystem.

How many known vulnerabilities does vLLM have?

vLLM has 188 known CVEs, 13 of them critical, tracked from NVD and GitHub Advisory.

Which ecosystem is vLLM distributed in?

vLLM is distributed via the pip ecosystem and categorized as llm inference.

Where does the vLLM vulnerability data come from?

Vulnerability data is sourced from NVD and GitHub Advisory, enriched with CVSS, EPSS, exploit signals, and patch status for each CVE.

How do I assess the risk of vLLM?

Review each CVE below — every entry shows CVSS severity, EPSS exploit probability, exploitation signals, and whether a patched version is available.

Monitor vLLM in your stack

Get instant alerts when new vulnerabilities affect vLLM. CISO analysis, ATLAS technique mappings, and compliance reports included.

Start Monitoring