An inference engine, model server, desktop app, browser workspace, and document platform are not substitutes for one another. Start with the layer you actually need.
Choose by job
Pick the layer you actually need: a graphical first setup, a local API, a browser interface, a document workspace, or a lower-level inference engine.
Best when you want model discovery, loading, and chat without assembling a stack.
Desktop app
The easiest first choice for a graphical desktop setup, model discovery, and local chat.
Desktop app
A desktop alternative when you want local chat plus optional provider connections in one app.
Desktop app
Consider it for a desktop chat and LocalDocs workflow, but compare the current product state against newer alternatives first.
Best when another app or your own code needs a local model service.
Model runner or API
The clearest first choice for a lightweight local model runner, command-line workflow, or API used by other apps.
Best when you want a web interface over one or more model providers.
Browser workspace
Use it as a browser interface after the underlying model provider already works.
Best when ingestion, embeddings, retrieval, and document organization are part of the job.
Documents and RAG
Use it when the main job is organizing documents and retrieval, not merely opening a chat window.
Best when you need backend control and are comfortable assembling the surrounding workflow.
Inference engine
Use it when you need low-level control over GGUF inference, backends, command-line options, or an embeddable server.