The FAR AI’s Model Registry is a decentralized marketplace and orchestration layer for AI models. It allows nodes to dynamically host and serve models tailored to their hardware capabilities, while developers and clients can select the optimal inference configuration for their use case.
When a node joins the network, it installs the FAR Launcher, which performs several critical tasks automatically:
Hardware Assessment: The launcher scans the node’s GPU VRAM and bandwidth to determine which models can run most efficiently on the available hardware.
Optimal Model Selection: Based on the hardware profile, the launcher downloads the model that maximizes node profitability.
Continuous Updates: The launcher monitors the registry for updated versions of open-source models. Whenever a new version is released, nodes automatically update their weights and binaries.
Encrypted & Verified Distribution: All models are distributed through a secure, internal registry with cryptographic verification to ensure authenticity.
