Local & Offline

Run models on your machine. No card, no vendor quota, works when the line drops.

Desktop app is free for individuals. Models are downloaded separately (GGUF).

The GUI for local LLMs. OpenAI-compatible local server so Cline/OpenCode can talk to a model on your disk.

MIT. The runtime under LM Studio, Ollama, and most GGUF tools.

C/C++ LLM inference. Use the CLI when you want fewer moving parts than a desktop app.

GPL-3 graph UI for diffusion. Unlimited on your GPU. Models extra.

Node graph for local image (and now video) generation. The serious alternative to clicking Leonardo every day.

Open-source ChatGPT-style desktop. Models optional, fully local.

Local chat UI with an OpenAI-compatible server. Cleaner than raw llama.cpp if you just want to talk to a 7B.

MIT desktop chat. One-click model installer. Fully offline after download.

The onboarding ramp to local LLMs. Less power-user than LM Studio, easier for a first install.