What people run on Functions
Auto Endpoints is the fastest path from an open-weight model to an autoscaling HTTPS endpoint.
Isolated, ephemeral containers started from code, for coding agents, background agents, and RL rollout pipelines.
Single-GPU fine-tuning through multi-node runs, without managing clusters or reserving capacity.
Evals, embeddings, re-ranking, and dataset generation as thousands of parallel jobs from one call, with no orchestration to manage.
Any of the runnable examples fit inside the monthly credits.
|