MIT

Run open LLMs (Llama, Mistral, Gemma, Qwen) locally with one command.

Framework for building LLM apps — prompts, chains, agents, tool use, RAG.

Self-hosted, polished web UI for Ollama and any OpenAI-compatible LLM.

Multi-user chat over your documents — works with any local or hosted LLM.

MIT-licensed open-source automation platform — proper OSS Zapier alternative.

Tiny C/C++ inference engine for LLaMA-family models — runs on CPU + many GPUs.

OpenAI-compatible REST API for local models — drop-in replacement for openai.com.

Data framework for connecting LLMs to your private data (the RAG specialist).

OpenAI Whisper speech-to-text, ported to C/C++ — fast, offline transcription.

Fast, small neural TTS that runs on a Raspberry Pi — by the Rhasspy project.