Model Workbench Tutorial: Choose, Run, and Save Results from /models

2026-11-12 · 原创·功能教程

Model Workbench: Choose, Run, and Save

The Model Workbench is a practical starting point for readers who are new to AI models, as well as for users who need to tune parameters or save media results. The workflow is straightforward: choose a model, review its details, run it with your input, inspect the result, copy example code that matches the current input, and save supported media results to your private results library. The verified entry point is `/models`.

Before you start

Define the task you want to complete and prepare the content you plan to enter. Open `/models` to browse the model catalog. If you are unsure which model to choose, open a specific model detail page and compare the displayed capabilities with your task. Use the information shown on the page rather than relying on external memory for model parameters or pricing.

Step-by-step operation

1. Enter the model catalog from `/models` and open a specific model detail page.

2. Review capabilities, pricing, and the model ID. Capabilities help you judge whether the model fits the task; pricing lets you check cost information before running; the model ID identifies the model.

3. Enter your content as needed and run the model. The example code follows the current input, so it is not a fixed snippet unrelated to your task.

4. Inspect the result. If the output is a supported media result, you can save it to your own results library.

5. Copy the example code matched to the current input. Before using it, check the model ID and the input again.

Reviewing and saving results

First check whether the output meets your task requirements, then decide whether to keep it. Supported media results can be saved privately in your results library. Because example code changes with the current input, copy it again after you change the input. This helps keep the code and the run consistent.

Safety and compliance notes

Do not submit sensitive information that is not permitted for processing. Before running a model, review the displayed capabilities and pricing information on the model page, and avoid filling in numbers from memory. When saving media results, consider how the content will be used and managed in your results library.

After completing these steps, return to the model catalog to compare capabilities, pricing, and model IDs across models. You can then choose a more suitable model and repeat the same workflow for another trial.