Choosing Models

Choose models by task quality, cost, speed, data handling and availability.

In this lesson you will

Compare hosted and open models
Select models against explicit criteria

Key concepts

hosted vs open-weight modelsmodel size and qualitydata handling and locationvendor dependence

The lesson

Hosted models are accessed through providers' APIs: strong and easy to start with, but data leaves your systems and costs are paid in foreign currency. Open-weight models can be run on your own hardware, keeping data local, but need technical capacity.

Choose against explicit criteria measured on your own evaluation set: quality, cost per task, speed, languages supported, data policies and availability.

Avoid lock-in: keep prompts and evaluation sets provider-neutral and design so models can be swapped.

Step by step

List candidate hosted and open models.
Run the evaluation set on two or three.
Compare against the criteria.
Choose and document a fallback.

Practice activity

Select the capstone model with evidence and a fallback option.

Expected result: A documented choice with evaluation results and a fallback.

Check your understanding

What is a key advantage of open-weight models?

Show answer
They can run on your own hardware, keeping data local.
How can lock-in to one provider be reduced?

Show answer
Keep prompts and evaluations provider-neutral and design for swappable models.

Taught live by an instructor — online or in person. Free course lessons are for self-study and do not lead to a certificate — see registration options.