Local and Self-Hosted AI

Self-hosted personal AI assistant on your own VPS that lives in Telegram, with persistent long-term memory, voice, and Claude-powered reasoning.

Open-source local chat application that runs LLMs on consumer hardware without cloud dependencies.

Open-source ChatGPT alternative that runs 100% offline with local model management and privacy guarantees.

Distributes LLMs as single executable files with zero setup required from Mozilla.

C/C++ LLM inference engine supporting CPU, GPU, and Apple Silicon as the foundation of local AI.

Desktop application for running local LLMs with a polished UI across all major platforms.

Drop-in OpenAI API replacement that runs models locally without requiring a GPU.

Run LLMs locally with a dead-simple CLI interface and 162K+ GitHub stars.

Self-hosted Telegram AI assistant written in Rust with sandboxed tool execution, MCP integration, and multi-agent orchestration.

High-throughput LLM serving engine with PagedAttention for production-grade local inference.

Local-first AI execution platform with Docker sandbox isolation, BYOK model configuration, MCP support, 5-stage Pipeline, and multi-platform messaging via WeChat, Feishu, DingTalk, Telegram, and Discord.