CVE-2026-108836
Description
A flaw was found in Ollama. Improper input validation within the model name parsing logic allows a remote attacker to conduct a Server-Side Request Forgery (SSRF) attack through the pull API. By supplying a crafted model name, an attacker can coerce the server into sending unauthorized network requests to internal or restricted network endpoints, potentially exposing sensitive internal services.
Statement
This vulnerability is rated as Important severity because an unauthenticated remote attacker can coerce the service into dispatching requests to arbitrary internal systems and cloud metadata endpoints without credentials. While default configurations typically restrict the listener to loopback interfaces, exposure occurs when the service is intentionally configured to listen on external interfaces or behind an unauthenticated proxy. Red Hat Enterprise Linux does not package or distribute Ollama; only community releases such as Fedora carry this package.
Frequently Asked Questions
Not sure what something means? Check out our Security Glossary.
Want to get errata notifications? Sign up here.
