LLMMedium
CVE-2025-48944: vLLM is an inference and serving engine for large language models (LLMs). In version 0.8.0 up to but…
- Occurred
- May 30, 2025
- Published
- May 30, 2025
- Source
- NVD CVE
Summary
vLLM is an inference and serving engine for large language models (LLMs). In version 0.8.0 up to but excluding 0.9.0, the vLLM backend used with the /v1/chat/completions OpenAPI endpoint fails to validate unexpected or malformed input in the "pattern" and "type" fields when the tools functionality is invoked. These inputs are not validated before being compiled or parsed, causing a crash of the inference worker with a single request. The worker will remain down until it is restarted. Version 0.9.0 fixes the issue.
Tags
- CVE
- CVE-2025-48944
- cvss:MEDIUM
Classification
Category confidence: 95%
Severity confidence: 90%
Rationale: vLLM LLM serving engine vulnerability causes worker crashes via malformed input, affecting availability but not causing physical harm or data breach.
Classifier: Qwen/Qwen3.5-35B-A3B-FP8