CVE-2026-44223

6.5 MEDIUM
Published: May 12, 2026 Modified: May 15, 2026
View on NVD

Description

vLLM is an inference and serving engine for large language models (LLMs). From to before 0.20.0, the extract_hidden_states speculative decoding proposer in vLLM returns a tensor with an incorrect shape after the first decode step, causing a RuntimeError that crashes the EngineCore process. The crash is triggered when any request in the batch uses sampling penalty parameters (repetition_penalty, frequency_penalty, or presence_penalty). A single request with a penalty parameter (e.g., "repetition_penalty": 1.1) is sufficient to crash the server. This vulnerability is fixed in 0.20.0.

AI Explanation

Get an AI-powered plain-language explanation of this vulnerability and remediation steps.

Login to generate AI explanation

CVSS v3.x Details

0.0 Low Medium High Critical 10.0
Vector String
CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H

References to Advisories, Solutions, and Tools

Patch Vendor Advisory Exploit Third Party Advisory
https://github.com/vllm-project/vllm/pull/38610
Source: security-advisories@github.com
Issue Tracking Patch
https://github.com/vllm-project/vllm/security/advisories/GHSA-83vm-p52w-f9pw
Source: security-advisories@github.com
Mitigation Vendor Advisory
https://github.com/vllm-project/vllm/pull/38610
Source: 134c704f-9b21-4f2e-91b3-4a467353bcc0
Issue Tracking Patch
https://github.com/vllm-project/vllm/security/advisories/GHSA-83vm-p52w-f9pw
Source: 134c704f-9b21-4f2e-91b3-4a467353bcc0
Mitigation Vendor Advisory

4 reference(s) from NVD

Quick Stats

CVSS v3 Score
6.5 / 10.0
EPSS (Exploit Probability)
0.4%
28th percentile
Exploitation Status
Not in CISA KEV

Weaknesses (CWE)

Affected Vendors

vllm