Tools for Local/Private AI

Run AI on your own machine. No cloud accounts, no data leaving your network, no monthly bills. Everything here is open source or has a fully local option.

See also: The Local Intelligence Stack — our recommended setup with getting-started guide.

Local Language Models

• Ollama — The easiest way to run LLMs locally. One-command install, simple CLI, and an API compatible with OpenAI’s format. Run Llama 3, Mistral, CodeLlama, and dozens more.
• LM Studio — GUI for running local models. Download, chat, and use as a local API server. Good if you prefer a visual interface over CLI.
• llama.cpp — The engine under Ollama. Use directly for maximum control and performance tuning.

Local Image Generation

• ComfyUI — Node-based UI for Stable Diffusion. Maximum control, custom workflows, no content filters. Requires a GPU (8GB+ VRAM minimum, 16GB+ recommended).
• Automatic1111 (Stable Diffusion WebUI) — Browser-based interface for Stable Diffusion. Slightly easier to start with than ComfyUI, fewer advanced features.
• Fooocus — Simplified Stable Diffusion interface. Minimal settings, good results. Best for people who want images without learning ComfyUI.

Local Speech

• Whisper (OpenAI) — State-of-the-art speech-to-text that runs locally. pip install openai-whisper and transcribe anything. Multiple model sizes for different speed/accuracy tradeoffs.
• Piper — Local text-to-speech. Not ElevenLabs quality, but completely free and offline. Good for accessibility and basic narration.
• Bark — Open-source text-to-speech with emotion and multilingual support. More natural than Piper, heavier on resources.

Local Knowledge & Notes

• Obsidian — Local-first markdown notes with a plugin ecosystem. The “Ollama” community plugin lets you chat with a local LLM inside your notes. Your data never leaves your disk.
• Logseq — Open-source, local-first outliner. Block-based references, daily journals, graph view.

Local Coding Assistants

• Continue.dev — Open-source AI coding assistant that works with VS Code. Point it at your local Ollama instance for fully private code completion.
• Tabby — Self-hosted AI coding assistant. Code completion backed by local models.

Hardware Requirements