Inference overview
Run text, image, video and decision models with one Inworld API key. Choose a model and use the guide for your task.
Inworld Inference gives you access to text, image, video and decision models with one API key. Select a model for your task, call it directly, and use the result in your application.
Choose a model
Browse inworld.ai/models to compare models and pricing and copy an API model ID. Then use the guide for the output you need. Each guide explains its request format and model-specific constraints.
| Task | Guide | How it works |
|---|---|---|
| Generate text or chat | Chat completions | Send messages and receive a model response. |
| Generate images | Image generation | Send a prompt or reference image, then save the generated image. |
| Generate video | Video generation · Image-to-video · Reference-to-video | Create a job, poll its status and download the video. |
| Classify, choose or score | Decision models | Send state and typed questions; receive answers your code can act on. |
| Read fields from text or images | Extraction models | Send content and typed fields; receive each value with its confidence. |
Make a request
Create an API key in Inworld Portal. Follow the Quickstart for a first text-generation request, including authentication and reading the response. For images, video, decision or extraction models, use the corresponding guide above.
You can call a model directly without creating a router. If your application needs fallbacks, see Request-level routing. For conditional routes or traffic splitting, see the Routing quickstart.