Species line up at a pier beside open bottles while a crane lowers another bottle toward the sea.

Service · Inference Nodes · Prisman

Compute Nodes for inference

Assign compute Nodes to model-serving workloads that provide inference to applications and Agents.

Place inference on compute capacity

Inference Nodes pair a model-serving workload with compute resources selected for its requirements. Node compute and inference service terms are accounted for separately.

Inference Nodes

Select the capacity for the service

Node profile, location, and inference terms shape the deployment for a model-serving workload.

Define the workload

Describe the model-serving use and its expected operating context.

Choose compute

Select an available Node profile for the assigned workload.

Connect inference

Expose the inference service to applications or Agents through its applicable API or contract.

Inference Nodes

Node resources support inference

Compute resources host the workload; Prisman’s inference service handles the model requests.

Compute profile

Choose Node resources according to the model-serving workload.

Location

Select an available location when contracting the Node.

Inference contract

Review API usage or custom contract terms separately from Node compute.

Species prepare beside open bottles in a sheltered coastal cove.

Service · Inference Nodes

Scope an inference deployment

Contact Prisman about compute capacity and inference terms for your model-serving workload.