Gaianet Node

Gaia is a framework for running a decentralized AI agent service on hardware you control, and a Gaia node is the unit that does the work. Installing one gives you a self hosted inference endpoint with retrieval built in: the node loads a language model, attaches a vector database holding your own knowledge, and serves both a chat interface and an API that other software can call. A node bundles several open source components. WasmEdge with its GGML plugin runs the model, Qdrant stores the embeddings that back the knowledge base, a LlamaEdge API server exposes the endpoints, a web dashboard provides the chat interface, and frpc handles public reachability so a node can be published at a public address or kept on localhost only. The default configuration ships a Llama 2 7B chat model with a MiniLM embedding model, and pointing the node at a different configuration URL swaps in other models and knowledge bases. Installation is a single script on macOS, Linux or Windows under WSL, after which gaianet init downloads the models and vector data, gaianet start brings the node up, gaianet stop shuts it down and gaianet config adjusts settings. The audience is developers and operators who want to run and customize their own agent rather than rent capacity from a hosted provider, and who want the knowledge base to stay on their own machine. The node software is distributed under the GPL 3.0 license from a public repository, so there is no license fee and the cost is the hardware it runs on. Gaia publishes no pricing page; the company describes the wider project as an ecosystem supporting applications that learn, improve and grow over time.