Run a model in Predictions before you write code
·5 min read·Model Market Product Team
Use the catalog and Predictions to validate a prompt and price before you integrate the same model with an API key.
Start from the catalog
Open AI Models and filter by task type — text-to-image, image-to-video, and so on. Each card shows example output, unit price, and recent activity so you can compare before you spend credits.
Try it in Predictions
On a model page, fill the input schema and run. You see the same job shape your API call will use: inputs in, files or text out, with status and cost on the run history under Predictions.
Promote the same model to your app
When the output looks right, create an API key under Account and call the model with the same identifier. Swap the model string later without rewriting the rest of your integration.
Quick tips
- Check unit price on the model card before a long test session.
- Confirm your email before buying credit.
- Use Favourites to keep a shortlist of models you are evaluating.