← Back to search

CVE-2026-94625

5.3 MEDIUMpublic exploit available

Published 2026-09-21 · Updated 2026-09-21

AI risk analysis

Summary
The vulnerability allows attackers to send rejected prefill requests that create ownerless transfer placeholders, leading to resource exhaustion and delayed processing of valid requests.
Exploitability
Exploitation requires sending specific rejected requests; it is moderately difficult due to the need for understanding the system's request handling mechanisms.
Blast radius
If exploited, this could significantly impact service availability by delaying or blocking valid requests, affecting user experience and potentially leading to operational disruptions.
Prioritized remediation
Update vLLM to version 0.29.1 or later, which addresses this vulnerability according to the project's security updates.
resource-exhaustiondelayvulnerabilityupdate

Analysis generated locally by qwen2.5:7b-instruct (no data left the box). AI-assisted — verify against primary sources before acting.

NVD description

vLLM through 0.29.0 contains a resource exhaustion vulnerability in MooncakeConnector where rejected prefill requests create ownerless transfer placeholders that are never reclaimed. Attackers can send rejected requests to exhaust sender task pools, causing valid requests to be delayed by up to 480 seconds while health checks continue returning success.

CVSS vector

CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:L

Weaknesses

CWE-772

Public exploit & PoC references

All references

Source data: NVD (nvd.nist.gov), public domain. Exploit-DB.ai adds local AI analysis for defensive use only.

Related CVEs

Related by shared AI tags and CWE weakness class. Browse the full archive.