Hyperspace
Decentralized P2P AI inference network
Performances
RD Popularity Index
RD Growth Index
Comparison
Fundraising
Follow Updates
Follow Lists
People
About Hyperspace
Hyperspace is a peer-to-peer decentralised AI inference and search network that also runs an AI-native chain; it is not a centralised AI search product. Each node handles crawling, indexing, ranking and inference at once, and queries resolve in stages from local cache to peers to a distributed hash table to live crawling; a proof mechanism verifies experimental execution with a zero-knowledge virtual machine and posts it on-chain. The pain point is that centralised inference and search concentrate compute cost and query logs and allow single-point censorship.
In March 2026 it launched a distributed artificial general intelligence agent network, in April it reported 660 active agents, 27,247 experiments and chain version 1.5.7, and in July it published a dual-swarm experiment reproducing an external model event on its own network. The network, command-line tool and chain all run, and the search subsystem offers a node installer.
Hyperspace advanced its AI-agent infrastructure in 2026. In April 2026 it launched 8 subdomains, HyperSearch and agent infrastructure, and upgraded the AVM to a CLI sidecar v5.40.0 (April 29). Around the same time it introduced the Gossiping Agents Protocol, shipping a TypeScript SDK v2.0.4, a Python SDK and a reference implementation so developers can quickly build decentralized agent networks that communicate with each other.
Looking ahead, Hyperspace plans to expand the protocol's multi-agent collaboration and decentralized indexing, integrate AVM and HyperSearch into more Web3 apps, and build an open agent ecosystem and developer toolchain for the AI x crypto space.
Varun Mathur — Co-founder & CEO; former VP of data at a Wall Street firm, previously built two products used in 100+ countries.