Libertai

LibertAI sells model inference that keeps no logs. The API stays OpenAI compatible, so a developer changes one base URL to move an existing application over, and models run on Aleph Cloud's distributed nodes, optionally inside a trusted execution environment where the operator cannot read prompts or outputs.

Beyond text and vision models the platform offers knowledge bases, image generation and an agent runtime: an agent holds its own wallet, pays per call in USDC through the x402 protocol and tops up compute credit on Aleph, so no API key or billing dashboard is needed, and deployment means pushing a container to a decentralised virtual machine. In November 2024 the team demonstrated on-chain agents running on Base and Solana at Devcon, and by early 2025 cumulative text completions had passed 1.7 million.