Ollama desktop client

Use Ollama models through the Glyph workspace.

Ollama provides a local model runtime and API. Glyph adds a desktop workspace around supported model endpoints so users can work with chat, knowledge, and other Glyph tools while administrators retain control over approved destinations.

Download Glyph

How the pieces fit

Ollama owns model storage and inference. Glyph connects to the configured endpoint and presents the user workflow. Keeping the endpoint on loopback confines access to the same computer; exposing it on a network requires TLS, authentication, and network controls appropriate to that environment.

Choose and test a model

Pull a model whose license and memory needs fit the machine, confirm it responds through Ollama, then register or select it in Glyph. Test context length, tool behavior, latency, and output quality with representative tasks before making it an organizational default.

Private by configuration

A local Ollama endpoint can keep inference on owned hardware. Glyph features that deliberately use web search, hosted services, or external model providers still follow those destinations. Review the selected provider and policy rather than assuming one local integration governs every request.