DaiHive: P2P network that runs one big local LLM across several PCs
DaiHive (daihive.eu) is a P2P () network for running without needing one powerful machine. A large model gets split across multiple worker computers on the network, and the user prompts it just like a normal single model.
Besides joining the public pool, people can spin up their own private, isolated team network — for example, pooling around 10 office PCs to run a model too big for any single machine, without exposing it to the wider network. The project is still in beta, with frequent updates to both the worker and the UI.
The range of supported model families is currently limited, but the system is al today, not just a concept. The developer is looking for , on the worker and UI, and input on which model families to prioritize next.
Key points
- DaiHive is a P2P network that splits one large LLM across multiple machines to run it
- Besides the public pool, you can build a private, isolated team network from your own machines
- Still in beta with limited model support, but al and actively developed
- Developer is seeking and on which models to support next