Inference

Call open models with an OpenAI-style client. Token charges come out of the same credit as GPU time. Keys are created in Console.

Step 1

Create key

Create a scoped key in Console and keep it in your runtime secret manager.

Step 2

Pick model

Choose a model endpoint and context window from the pricing tables in docs.

Step 3

Ship safely

Set daily budget guards so agents can test fast without runaway spend.

Manage API keysOpen API docs