Models

Keep your models in one place, add them from Kaggle or from a local GGUF file, and run a local model right inside the editor.

What you can do

• Browse your models — each model shows up as a card with its name, source, task, context window, and modalities.
• Add from Kaggle — search Kaggle and import a model's details into your catalog. This adds the model's metadata, not its weights.
• Add a local model — pick a GGUF file you already have on disk and add it to your catalog.
• Run a local model — start a local llama.cpp server for a GGUF model with one click; the editor's chat and completions can then use it as their model.
• Remove — take a model out of your catalog (files on disk are left in place).

Requirements

• Add from Kaggle needs the Kaggle Models extension.
• Adding and running local models needs the desktop app (Electron) — picking files and running a local model server aren't available in the browser host.

Notes

• Kaggle imports bring in metadata only — no weights are downloaded. "Add a local model" expects a GGUF file you already have.

Getting started

Open the Models sidebar from the activity bar and press + to add your first model.