CVE-2026-73560

The record

vLLM is an inference and serving engine for large language models. Prior to 0.26.0, the MiMoV2OmniMultiModalProcessor in vllm/transformers_utils/processors/mimo_v2_omni.py passes attacker-controlled image and audio strings through _fetch_image, requests.get, and Image.open instead of MediaConnector, bypassing allowed_media_domains and allowed_local_media_path protections and allowing server-side requests and reads of arbitrary files accessible to the vLLM process. This issue is fixed in version 0.26.0.

Affected packages

Related vulnerabilities

• CVE-2026-114730 Roundcube Webmail 1.6.14 before 1.6.20 and 1.7.0 before 1.7.5 is affected by a… CWE-918
• CVE-2026-87953 Server-Side Request Forgery (SSRF) vulnerability in Drupal SAML SSO - Service Provider… CWE-918
• CVE-2026-114634 WWBN AVideo before 29.4.0 contains a blind server-side request forgery vulnerability in… CWE-918
• CVE-2026-112288 Krayin CRM 2.2.5 before 2.2.7 contains a server-side request forgery filter bypass that… CWE-918
• CVE-2026-111936 Sigma File Manager through 2.2.0 contains a server-side request forgery vulnerability… CWE-918
• CVE-2026-94483 Next.js is a React framework for building full-stack web applications. CWE-918
• CVE-2026-102267 PyJWT is a Python implementation of JSON Web Token standards. CWE-918
• CVE-2026-101913 ip-address is a library for parsing and manipulating IPv4 and IPv6 addresses in… CWE-918