examples/

Command chat-completion runs a non-streaming chat completion against a DigitalOcean Agent Inference endpoint.

Command cancel-model-evaluation-run cancels an in-progress model evaluation run via the GradientAI API.

Command create-model-eval-dataset-upload-presigned-urls creates presigned URLs that can be used to upload model evaluation dataset files via the GradientAI API.

Command create-model-evaluation-run creates a new model evaluation run via the GradientAI API.

Command delete-model-evaluation-preset deletes a saved model evaluation preset via the GradientAI API.

Command get-model-evaluation-preset retrieves a saved model evaluation preset via the GradientAI API.

Command get-model-evaluation-run retrieves a model evaluation run, including the run summary, aggregated result metrics, and a paginated list of per-prompt results, via the GradientAI API.

Command get-model-evaluation-run-results-download-url returns a presigned download URL for a model evaluation run's results (gzip-compressed JSON) via the GradientAI API.

Command list-evaluation-datasets lists evaluation datasets via the GradientAI API.

Command list-model-evaluation-metrics lists all available metrics that can be selected when creating a model evaluation run, via the GradientAI API.

Command list-model-evaluation-presets lists all saved model evaluation presets via the GradientAI API.

Command list-model-evaluation-runs lists model evaluation runs, optionally filtered by preset UUID, status (single or multiple), candidate model source types, and free-text search.

Command chat-completion runs a non-streaming chat completion via the Serverless Inference API.

Command chat-completion-streaming streams a chat completion via the Serverless Inference API.

Command embeddings creates embedding vectors via the Serverless Inference API.

Command image-generation generates a single image via the Serverless Inference API and writes it to image.png.

Command image-streaming streams partial-image events via the Serverless Inference API.

Command list-models lists the models reachable through the Serverless Inference API.

Command messages sends a non-streaming message via the Anthropic-compatible /v1/messages endpoint of the Serverless Inference API.

Command messages-streaming streams an Anthropic-style message from the Serverless Inference /v1/messages endpoint.

Command responses runs a non-streaming /v1/responses call via the Serverless Inference API.

Command responses-streaming streams a /v1/responses call via the Serverless Inference API.