Skip to main content
Brightnode offers dedicated compute, hosted inference, and shared storage for AI workloads. The current public product surface is built around Router, Models, Beams, Bnodes, and Network volumes.
Some older docs and examples may still mention Serverless, Hub, or Public Endpoints. In the current product, those ideas map most closely to Router, Models, and Beams.

​
Router

Router gives you a single OpenAI-compatible endpoint for Brightnode-hosted inference. You authenticate with a Brightnode API key, choose a model ID, and use standard OpenAI SDKs or raw HTTP calls.

​
Models

Models is the catalog of model IDs available through the hosted inference API. It helps you discover pricing, context length, modality, provider health, and quickstart examples before you send traffic.

​
Beams

Beams is the preview path for managed model deployments. It is intended for cases where you want dedicated inference capacity and more control over deployment behavior than the shared router provides.

​
Bnodes

Bnodes give you dedicated GPU or CPU machines with full control over the environment. They are the right fit for training, experiments, notebooks, custom containers, and long-running workloads.

​
Network volumes

Network volumes provide portable persistent storage for datasets, checkpoints, and shared assets. They work especially well with Bnodes and can also be managed directly through the S3-compatible API.

​
Choosing the right option

Choose Router when you want the fastest path to production inference with a single API surface and minimal operational work. Choose Models when you are deciding which hosted model to call through Router and need to compare metadata, context length, pricing, or modality. Choose Beams when you need managed model deployments and are comfortable working with a preview product. Choose Bnodes when you need full machine control, custom images, or persistent interactive environments. Choose Network volumes when you need portable shared storage that survives the lifecycle of individual machines.