cvedb.io
CVE-2026-44223
MEDIUM · CVSS 6.5
EPSS exploitation probability: 0%
Published 2026-05-12T20:16:43.293 · Last modified 2026-06-22T22:16:45.507

Summary

vLLM is an inference and serving engine for large language models (LLMs). From 0.18.0 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 — vllm

Does this affect you?

Add your gear to cvedb and we'll alert you only when vllm ships something exploited.

Check my exposure →

References

This product uses data from the NVD API but is not endorsed or certified by the NVD. Informational only; not professional security advice.