CVE-2026-44223: medium-severity vulnerability in vllm-project vllm
vLLM: extract_hidden_states speculative decoding crashes server on any request with penalty parameters
Published · Updated
13Vexday Risk Score
No sign of exploitation. No public exploitation artifact known so far.
ssvc Trackcvss 6.5epss 0.4%
exploitation probability
0.4%top 65% of all CVEs
observed exploitation
nono source reports it
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.
CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H
Affected products
vllm-project · vllmRelated CVEs — vllm-project vllm
In the same product, most dangerous first.
CVE-2026-22778CRITICALvLLM leaks a heap address when PIL throws an errorEPSS 11.2%CVE-2026-27893HIGHvLLM's hardcoded trust_remote_code=True in NemotronVL and KimiK25 bypasses user security opt-outEPSS 1.8%CVE-2025-32444CRITICALvLLM Vulnerable to Remote Code Execution via Mooncake IntegrationEPSS 1.8%CVE-2026-48746CRITICALvLLM: OpenAI auth bypassEPSS 1.2%CVE-2025-47277CRITICALvLLM Allows Remote Code Execution via PyNcclPipe Communication ServiceEPSS 1.0%CVE-2025-62164HIGHVLLM deserialization vulnerability leading to DoS and potential RCEEPSS 0.9%