Decentralised AI Training At Scale

Our mission is to democratize AI by making model training accessible, enabling developers and data holders to make composable contributions of compute, training algorithms, and data.

Self-sustained data and model flywheel

Try AI Arena

Training node: Compete with other ML engineers to submit the best-performing models for a given training/ fine-tuning task

Validators: Score model submissions to ensure a fair way of distributing token rewards

Delegators: Stake to validators to amplify their influence

Try FL Alliance

FL node: Collaborate with other nodes to finetune models using proprietary data and local compute. Train models without handing over data ownership