CVE-2026-56340: vLLM introduces enhanced protection for sparse tensor validation DoS

# CVE-2026-56340: vLLM introduces enhanced protection for sparse tensor validation DoS

https://gcdradar.vercel.app/incidents/cve-2026-56340-vllm-introduces-enhanced-protection-for-spars · AIR-2026-0429 · closed 2026-01-08

- Lane: Patch now
- Severity: 2 (Disclosed vulnerability)
- Fix status: Fix status unknown
- First reported: 2026-01-08
- AI role: Attack on AI
- Products: vLLM
- CVE ids: CVE-2026-56340
- CVSS: 8.8

## Summary

An incomplete previous fix left vLLM vulnerable to a denial of service attack via missing sparse tensor validation when prompt embeds are enabled. The project is implementing additional checks to validate tensor indices.

## What to do

No action recorded; read the sources below and the vendor's advisory.

## Latest updates

- 2026-06-20: CVE-2026-56340: vLLM versions >= 0.10.2 and < 0.13.0 are missing sparse tensor validation in multimodal embeddings processing. (https://nvd.nist.gov/vuln/detail/CVE-2026-56340)
- 2026-01-08: CVE-2026-56340: vLLM introduced enhanced protection for CVE-2025-62164 (https://github.com/advisories/GHSA-mcmc-2m55-j8jj)

## Sources

- 2026-01-08 GitHub Advisory Database: CVE-2026-56340: vLLM introduced enhanced protection for CVE-2025-62164 https://github.com/advisories/GHSA-mcmc-2m55-j8jj
- 2026-06-20 NVD: CVE-2026-56340: vLLM versions >= 0.10.2 and < 0.13.0 are missing sparse tensor validation in multimodal embeddings processing. https://nvd.nist.gov/vuln/detail/CVE-2026-56340

Verify against the vendor's advisory before changing production. Data CC BY-SA 4.0, https://gcdradar.vercel.app.