Live › vulnerability › CVE-2026-112051
Go to dashboardXinference through 4.0.1 fails to block backslash sequences in model_name on Windows. An unauthenticated attacker can register a custom model named like ..\..\evil to write and delete arbitrary .json files the Xinference process can reach.
Summary written from the sources below. Check the originals before acting.
First seen by InfraStack 2026-10-23 21:57 UTC. All times on this page are UTC.
Teams running Xinference on Windows where the REST API is reachable, since POST /v1/model_registrations/{model_type} accepts model names with backslash sequences without authentication. Linux-only deployments are not described as affected by the source, and deployments that keep the API off untrusted networks are lower risk.
No product ranges from the sources yet.
xinference --version and compare against the fixed release listed in the advisory (unknown in sources)4.0.1 once a fixed release is published (fixed version unknown in sources)Each row is the first time that source carried this signal. The analyst’s rows are its own findings.
Xinference through 4.0.1 contains a path traversal vulnerability on Windows that allows unauthenticated attackers to write and delete arbitrary .json files via backslash sequences in model_name. Attackers can register a persisted custom model through POST /v1/model_registrations/{model_type} with names like ....\evil to overwrite configuration or data files writable by the Xinference process.
This page as .md · .json · in your agent: get_signal("CVE-2026-112051")
NVD data is provided by NIST and is public domain. Posts quoted belong to their authors.