What is LibertAI?
LibertAI is a decentralized AI platform that runs open-source models on Aleph Cloud, a distributed compute network with no central operator. You get the same developer experience as proprietary APIs — OpenAI-compatible, Anthropic-compatible — without sending your data through a single corporate gateway.
What you can build with it
| Chat & reasoning | Talk to open-source LLMs (Qwen, Gemma, DeepSeek, Hermes) via OpenAI or Anthropic SDKs — function calling, vision, and reasoning included. → Text API |
| Image generation | Stable Diffusion WebUI and OpenAI-compatible image endpoints. → Image API |
| Web search & retrieval | Multi-engine search (Google, Bing, DuckDuckGo, Brave, Semantic Scholar) with consensus ranking, plus a URL-fetch endpoint for RAG. → Search API |
| Autonomous agents | Build agents that pay for their own inference and compute with crypto — no API key, no billing dashboard. → Agents |
You can also use LibertAI through the chat interface or LiberClaw, our sovereign assistant frontend.
How it's different
- Decentralized infrastructure. Inference runs on independent compute nodes (CRNs) coordinated by Aleph Cloud, not a single provider's data center. Read the Architecture page for the request flow.
- Privacy options. Models marked with 🔒 run inside a Trusted Execution Environment so the operator running the hardware cannot see prompts or completions. See Trust model & TEE.
- Pay how you want. Use a traditional API key from the Developer console, or pay per request in USDC on Base via x402 payments — useful for agents that fund themselves.
- Open weights. Every model is open source (most are on Hugging Face). No black-box endpoints.
Where to start
- Got 5 minutes? → Quickstart — make your first call with curl, Python, or TypeScript.
- Coming from OpenAI? → Text usage — swap your base URL and keep your code.
- Building an agent? → Agents — Getting started.
- Want to understand the trust model? → Architecture and Trust model & TEE.