Decentralized AI races to revolutionize drug discovery with live competitions

A decentralised network is speeding up drug discovery by running three live competitions on Bittensor’s Subnet 68. Miners and validators are working together to find new treatments, using a system called Yuma Consensus to judge the best results. The approach promises faster and cheaper research than traditional labs. The first competition focuses on small molecule screening. So far, over 11 million molecules have been tested across nine disease targets. Miners submit potential drug candidates, while validators assess their quality through Yuma Consensus.

The second challenge targets nanobody design, specifically for PD-L1—a protein linked to cancer. Around 4,200 nanobody structures are currently under review. Validators use stake-weighted agreement to reward the most promising submissions. The third competition aims to improve search methods in chemical space. A total of 63 unique algorithms are competing to find the most efficient ways to explore possible drug compounds. Yuma Consensus ensures only high-quality outputs receive incentives. Bittensor’s decentralised model allows rapid screening of millions of candidates at a fraction of the usual cost. Metanova Labs is using this system to accelerate research while cutting operational expenses. The same framework also supports other tasks, like AI language model training.

The three competitions on Subnet 68 demonstrate how decentralised networks can transform drug discovery. By combining miner contributions with validator oversight, the system delivers faster, lower-cost research. Results from these challenges could lead to new treatments for diseases currently under investigation.