Medium severity6.5NVD Advisory· Published Aug 17, 2026· Updated Sep 18, 2026
CVE-2026-73560
CVE-2026-73560
Description
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.
AI Insight
LLM-synthesized narrative grounded in this CVE's description and references.
Affected packages
Versions sourced from the GitHub Security Advisory.
| Package | Affected versions | Patched versions |
|---|---|---|
vllmPyPI | < 0.26.0 | 0.26.0 |
Affected products
1Patches
Vulnerability mechanics
References
6- github.com/advisories/GHSA-4hhp-h66f-j5j7ghsaADVISORY
- nvd.nist.gov/vuln/detail/CVE-2026-73560ghsaADVISORY
- github.com/vllm-project/vllm/commit/54503ecec0f3ac31e5ecfc5f28652e4cc42307b5nvdWEB
- github.com/vllm-project/vllm/pull/43117nvdWEB
- github.com/vllm-project/vllm/releases/tag/v0.26.0nvdWEB
- github.com/vllm-project/vllm/security/advisories/GHSA-4hhp-h66f-j5j7nvdWEB
News mentions
0No linked articles in our index yet.