Loading...
Loading
Loading
Bet on
the algorithm.
Every Bittensor subnet is a different AI thesis. Pick the ones that produce real output. Stake TAO hard. Let the validators sort it out.
A blockchain where the mining output is actual AI.
Bittensor miners train models, run inference, and solve real problems. The compute goes to AI, not wasted on hash puzzles.
Thousands of miners and validators run across a global network. Anyone can join. Nobody owns it.
Each subnet is a different AI project. Put your TAO behind the ones you think will win and earn rewards when they do.
Staking is only the beginning. You can also contribute directly.
Stake to a root validator and compare its fund’s share price against the network index.
Browse funds →Run AI workloads on a subnet and earn TAO emissions for the compute you contribute.
Start mining →Define an AI task, write the incentive mechanism, and let miners compete on it.
Build a subnet →Score miner outputs, set weights on a subnet, and earn emissions for honest evaluation.
Learn to validate →Ready to get started?
The basics of Bittensor, TAO, and how the network turns compute into intelligence.
Bittensor is a decentralized network where miners produce AI models and services. Validators evaluate their output, and the network rewards contributors through its incentive system. Participants can contribute as miners, validators, subnet creators, or stakers. Bittensor fundamentals →
TAO is Bittensor’s native token, used for staking and transactions on the network. On non-root subnets, staking exchanges TAO for that subnet’s alpha token. The value of alpha relative to TAO can change. TAO and alpha tokens →
A subnet is a specialized network within Bittensor. Each defines a task, an incentive mechanism, and how validators evaluate miners’ work. Subnets can focus on services such as text generation, data collection, or computing. How subnets work →
Every root validator runs a fund. Stake to a validator and your root stake accrues shares of its fund; compare share price against the network index.