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 6 of 8)

SeverityCVE IDSummaryCVSSPublished
MEDIUMCVE-2026-54235vLLM: NaN/Inf bypass crashes GPU inference workers6.5Jun 17, 2026MEDIUMGHSA-8jr5-v98p-w75mvllm: EXIF/tRNS preprocessing gap enables adversarial input4.8Jun 17, 2026HIGHCVE-2026-53923vLLM: integer truncation leaks GPU memory cross-tenant7.5Jun 17, 2026MEDIUMCVE-2026-54236vLLM: heap address leak enables ASLR bypass5.3Jun 17, 2026MEDIUMCVE-2026-54233vLLM: decompression bomb OOM via audio endpoint6.5Jun 17, 2026MEDIUMCVE-2026-12491vLLM: image metadata mishandling corrupts multimodal inputs4.8Jun 17, 2026HIGHCVE-2026-41523vLLM: assert bypass → RCE via poisoned HuggingFace model7.5Jun 16, 2026CRITICALCVE-2026-48746vllm: auth bypass exposes OpenAI inference API9.1Jun 16, 2026HIGHCVE-2026-5201gdk-pixbuf: JPEG heap overflow crashes vLLM inference7.5Mar 31, 2026HIGHCVE-2026-4111libarchive: infinite loop DoS in RAR5 decompression7.5Mar 13, 2026MEDIUMCVE-2025-14831GnuTLS: TLS cert parsing DoS hits vllm inference5.3Feb 9, 2026HIGHCVE-2023-52356libtiff: heap overflow DoS in vLLM inference via TIFF input7.5Jan 25, 2024HIGHCVE-2026-5121libarchive: integer overflow in zisofs hits vllm containers7.5Mar 30, 2026HIGHCVE-2026-4424libarchive: RAR heap OOB read leaks memory in vLLM stacks7.5Mar 19, 2026LOWCVE-2026-10813LMCache: weak hash enables KV cache integrity bypass3.6Jun 4, 2026HIGHCVE-2026-5497vLLM: unauthenticated OOM DoS via video frame parsing7.5Jun 11, 2026MEDIUMCVE-2026-47155vLLM: revision pin bypass loads unreviewed artifacts6.5Jun 10, 2026UNKNOWNCVE-2026-4944vllm: trust_remote_code bypass enables RCE via HuggingFace--May 28, 2026MEDIUMCVE-2026-9540vllm: unauthenticated DoS in OpenAI-compatible serving path5.3May 26, 2026MEDIUMCVE-2026-44223vLLM: speculative decoding DoS via penalty params6.5May 6, 2026MEDIUMCVE-2026-44222vLLM: token injection DoS via multimodal placeholders6.5May 5, 2026MEDIUMCVE-2026-7141vllm: uninitialized KV cache memory leaks inference data5.6Apr 27, 2026MEDIUMCVE-2026-34753vLLM: SSRF in batch API exposes cloud metadata endpoints5.4Apr 3, 2026MEDIUMCVE-2026-34755vLLM: OOM DoS via unbounded video frame decoding6.5Apr 3, 2026MEDIUMCVE-2026-34756vLLM: DoS via unbounded n parameter causes OOM crash6.5Apr 3, 2026

Showing 126–150 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