CVE-2026-25960

vLLM is an inference and serving engine for large language models (LLMs). The SSRF protection fix for CVE-2026-24779 add in 0.15.1 can be bypassed in the load_from_url_async method due to inconsistent URL parsing behavior between the validation layer and the actual HTTP client. The SSRF fix uses urllib3.util.parse_url() to validate and extract the hostname from user-provided URLs. However, load_from_url_async uses aiohttp for making the actual HTTP requests, and aiohttp internally uses the yarl library for URL parsing. This vulnerability in 0.17.0.

CVSS v3 7.1 HIGH

7.1^/10

CVSS v3 : HIGH

Vector :

Exploitability : 2.8 / Impact : 4.2

Attack Vector NETWORK

Attack Complexity LOW

Privileges Required LOW

User Interaction NONE

Confidentiality Impact HIGH

Integrity Impact NONE

Availability Impact LOW

Scope UNCHANGED

References

Link	Resource
https://github.com/vllm-project/vllm/commit/6f3b2047abd4a748e3db4a68543f8221358002c0	Patch
https://github.com/vllm-project/vllm/pull/34743	Issue Tracking Patch
https://github.com/vllm-project/vllm/security/advisories/GHSA-qh4c-xf7m-gxfc	Not Applicable
https://github.com/vllm-project/vllm/security/advisories/GHSA-v359-jj2v-j536	Exploit Patch Vendor Advisory

Configurations

Configuration 1 (hide)

cpe:2.3:a:vllm:vllm:*:*:*:*:*:*:*:*

History

No history.

Information

Published : 2026-03-09 21:16

Updated : 2026-03-18 18:36

NVD link : CVE-2026-25960

Mitre link : CVE-2026-25960

CVE.ORG link : CVE-2026-25960

JSON object : View

Products Affected

vllm

vllm

CWE

CWE-918

Server-Side Request Forgery (SSRF)

{"id": "CVE-2026-25960", "cveTags": [], "metrics": {"cvssMetricV31": [{"type": "Secondary", "source": "security-advisories@github.com", "cvssData": {"scope": "UNCHANGED", "version": "3.1", "baseScore": 7.1, "attackVector": "NETWORK", "baseSeverity": "HIGH", "vectorString": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:L", "integrityImpact": "NONE", "userInteraction": "NONE", "attackComplexity": "LOW", "availabilityImpact": "LOW", "privilegesRequired": "LOW", "confidentialityImpact": "HIGH"}, "impactScore": 4.2, "exploitabilityScore": 2.8}, {"type": "Primary", "source": "nvd@nist.gov", "cvssData": {"scope": "UNCHANGED", "version": "3.1", "baseScore": 9.8, "attackVector": "NETWORK", "baseSeverity": "CRITICAL", "vectorString": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H", "integrityImpact": "HIGH", "userInteraction": "NONE", "attackComplexity": "LOW", "availabilityImpact": "HIGH", "privilegesRequired": "NONE", "confidentialityImpact": "HIGH"}, "impactScore": 5.9, "exploitabilityScore": 3.9}]}, "published": "2026-03-09T21:16:15.537", "references": [{"url": "https://github.com/vllm-project/vllm/commit/6f3b2047abd4a748e3db4a68543f8221358002c0", "tags": ["Patch"], "source": "security-advisories@github.com"}, {"url": "https://github.com/vllm-project/vllm/pull/34743", "tags": ["Issue Tracking", "Patch"], "source": "security-advisories@github.com"}, {"url": "https://github.com/vllm-project/vllm/security/advisories/GHSA-qh4c-xf7m-gxfc", "tags": ["Not Applicable"], "source": "security-advisories@github.com"}, {"url": "https://github.com/vllm-project/vllm/security/advisories/GHSA-v359-jj2v-j536", "tags": ["Exploit", "Patch", "Vendor Advisory"], "source": "security-advisories@github.com"}], "vulnStatus": "Analyzed", "weaknesses": [{"type": "Primary", "source": "security-advisories@github.com", "description": [{"lang": "en", "value": "CWE-918"}]}], "descriptions": [{"lang": "en", "value": "vLLM is an inference and serving engine for large language models (LLMs). The SSRF protection fix for CVE-2026-24779 add in 0.15.1 can be bypassed in the load_from_url_async method due to inconsistent URL parsing behavior between the validation layer and the actual HTTP client. The SSRF fix uses urllib3.util.parse_url() to validate and extract the hostname from user-provided URLs. However, load_from_url_async uses aiohttp for making the actual HTTP requests, and aiohttp internally uses the yarl library for URL parsing. This vulnerability in 0.17.0."}, {"lang": "es", "value": "vLLM es un motor de inferencia y servicio para modelos de lenguaje grandes (LLMs). La correcci\u00f3n de protecci\u00f3n contra SSRF para CVE-2026-24779 a\u00f1adida en 0.15.1 puede ser eludida en el m\u00e9todo load_from_url_async debido a un comportamiento inconsistente de an\u00e1lisis de URL entre la capa de validaci\u00f3n y el cliente HTTP real. La correcci\u00f3n de SSRF utiliza urllib3.util.parse_url() para validar y extraer el nombre de host de las URL proporcionadas por el usuario. Sin embargo, load_from_url_async utiliza aiohttp para realizar las solicitudes HTTP reales, y aiohttp utiliza internamente la biblioteca yarl para el an\u00e1lisis de URL. Esta vulnerabilidad en 0.17.0."}], "lastModified": "2026-03-18T18:36:10.323", "configurations": [{"nodes": [{"negate": false, "cpeMatch": [{"criteria": "cpe:2.3:a:vllm:vllm:*:*:*:*:*:*:*:*", "vulnerable": true, "matchCriteriaId": "5A72A44F-F5E3-44C0-AB4B-A0B5B6797EFA", "versionEndExcluding": "0.17.0", "versionStartIncluding": "0.15.1"}], "operator": "OR"}]}], "sourceIdentifier": "security-advisories@github.com"}