# CVE-2026-44223

## Summary

- **CVE ID:** CVE-2026-44223
- **Severity:** MEDIUM
- **CVSS Score:** 6.5 (CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H)
- **CWE:** CWE-131, CWE-704
- **Published:** May 12, 2026
- **Last Modified:** Jun 22, 2026

## 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.

## Affected Products

- vllm-project — vllm (>= 0.18.0, < 0.20.0)

## References

- [CNA](https://github.com/vllm-project/vllm/security/advisories/GHSA-83vm-p52w-f9pw)
- [CNA](https://github.com/vllm-project/vllm/pull/38610)

## Exploitation Prediction (EPSS)

- **EPSS Score:** 0.37%
- **EPSS Percentile:** 29.9

---
_Exported from OnDuty AI Vulnerability Intelligence on 2026-09-10._