CVE-2025-48944: vllm-project vllm Improper Input Validation
Vulnerability summary
CVE-2025-48944 is a medium (CVSS 6.5) Improper Input Validation vulnerability in vllm-project vllm. EPSS estimates a 0.61% chance of exploitation in the next 30 days, higher than 48.7% of scored vulnerabilities. 1 public exploit is recorded.
Advisory lineage
Upstream 0 · Downstream 0 · Related 2Timeline
- 30-05-2025PublishedVulnerability first disclosed.
- 30-05-2025Last modifiedVulnerability information updated.
Description
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.
CVSS metrics
3 scores · 3 sources · 1 version- OSV · PypiEcosystem advisory · calculated6.5MediumCVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H
- NVDNIST6.5MediumCVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H
EPSS trend
Probability
0.61%
Chance of exploitation in the next 30 days
Percentile
49%
Versus all other CVEs
30-day change
+0.10 pp
From 0.52% on 21-09-2026
Known exploited
Compare KEV catalogsNot in any of the 3 tracked catalogs.
Public exploits
1 foundVllm
Maturity: poc
Weaknesses
CWE · root causeAffected systems
PyPI1 product
- vllm≥ 0.8.0, < 0.9.0
Vllm1 product
- vllm≥ 0.8.0, < 0.9.0
Vllm-project1 product
- vllm≥ 0.8.0, < 0.9.0
3 vendors · 3 products
References
7 links- github.com/vllm-project/vllm/security/advisories/GHSA-vrq3-r879-7m65X Refsource CONFIRMWEBExploitVendor Advisory
- github.com/vllm-project/vllmPACKAGE