Replicate

1. Running models

   Running a model by owner/name through client libraries and the HTTP API.

2. Predictions

   The prediction object as the unit of work, creation, inputs, and outputs.

3. Prediction lifecycle

   Statuses from starting through succeeded, failed, or canceled and polling.

4. Streaming output

   Server-sent event streaming of model output tokens.

5. Input and output files

   Passing files as URLs or data and retrieving generated output files.

6. Webhooks

   Receiving prediction updates via webhooks and verifying their signatures.

7. Models and versions

   Official, community, and private models, and model versions.

8. Deployments

   Dedicated, scalable endpoints with controlled hardware and scaling.

9. Pushing your own model with Cog

   Packaging a model with Cog and publishing it to Replicate.

10. Hardware

    GPU and CPU hardware options and their effect on cost and speed.

11. Rate limits

    Request limits per account and handling throttling.

12. API tokens and security

    Managing API tokens and keeping credentials safe.

Each section links to the official page. Titles and summaries by Extra Credit; the docs belong to their authors.