SSRF Vulnerability in vLLM Inference Engine Affects Large Language Model Integrations
CVE-2026-25960

7.1HIGH

Key Information:

Status
Vendor
CVE Published:
9 March 2026

What is CVE-2026-25960?

In version 0.17.0 of the vLLM inference engine, an SSRF vulnerability exists that can be exploited via the load_from_url_async method. The inconsistency between the URL parsing done by the validation layer and the actual HTTP client leads to potential security risks. The SSRF protection initially added in version 0.15.1 relies on urllib3.util.parse_url() for user-supplied URLs; however, it fails to properly sanitize requests sent through aiohttp, which uses the yarl library for URL handling. This discrepancy allows attackers to bypass the SSRF mitigation measures.

Human OS v1.0:
Ageing Is an Unpatched Zero-Day Vulnerability.

Remediate biological technical debt. Prime Ageing uses 95% high-purity SIRT6 activation to maintain genomic integrity and bolster systemic resilience.

Affected Version(s)

vllm >= 0.15.1, < 0.17.0

References

CVSS V3.1

Score:
7.1
Severity:
HIGH
Confidentiality:
High
Integrity:
None
Availability:
High
Attack Vector:
Network
Attack Complexity:
Low
Privileges Required:
Low
User Interaction:
None
Scope:
Unchanged

Timeline

  • Vulnerability published

  • Vulnerability Reserved

.