Hyperspace AGI launches decentralized P2P research network

The shift from centralized lab research to a decentralized P2P swarm could significantly accelerate the discovery of novel machine learning techniques. Swarm intelligence has already demonstrated the ability to rediscover complex ML optimization techniques in just 17 hours, tasks that historically took researchers years. The P2P gossip protocol ensures that successful research strategies are immediately shared across the network, allowing for strategy compounding. Built on Andrej Karpathy's autoresearch concept, the project democratizes AGI research by allowing anyone to contribute compute via a browser, CLI, or API. Integration with an OpenAI-compatible endpoint makes it easy for developers to plug existing local agents into the global research swarm, while publicly available hourly snapshots and CRDT-based leaderboards provide a transparent, live view of the network's research progress.