Choosing the Right Model for a Task
Pick a model by testing it on your own task, weighing quality, speed, cost, context and data handling.
A taste of a lesson
We are classifying support emails into 8 categories. Do we need the biggest model?
Probably not, but test rather than guess. Classification into a fixed set of labels is often handled well by smaller, faster models, especially with clear category definitions and a few examples. Take 50 to 100 real emails with agreed labels, run them through two or three models of different sizes with the same prompt, and compare accuracy, latency and cost per email. If a small model is within your quality bar, it wins. Who on your team can label the test emails?
Written by the teacher as an example. In your lesson the tutor answers your own questions, and like any AI it can be wrong.
What you will be able to do
- List the dimensions that matter when choosing a model for a feature
- Explain when a reasoning focused or a small fast model is the better fit
- Run a small comparison of shortlisted models on your own examples
- Estimate cost per successful answer, not just price per token
- Plan for model updates, deprecations and routing between models
Lesson plan
- 1 Requirements before models Write down what the feature needs before looking at any model list. Start
- 2 The dimensions of choice Compare models on quality, speed, cost, context, inputs and reliability. Start
- 3 Sizes, reasoning models and open weights Understand the main families of trade off between model types. Start
- 4 Testing on your own examples Run a fair comparison of a few shortlisted models. Start
- 5 Deciding and routing Choose a model or a routing scheme from the comparison results. Start
- 6 Living with model change Keep the choice healthy as models are updated and retired. Start
Try asking
About this tutor
For developers, product people and team leads deciding which model to use for a feature, and when to revisit that choice. You learn the dimensions that actually matter (quality on your task, latency, cost per request, context length, input types, tool use and structured output reliability, data handling and hosting options, open weight versus hosted), how reasoning focused models differ from fast general models, and how to run a small comparison on your own examples rather than trusting leaderboards. You leave with a repeatable selection process and a plan for model updates and retirements.
Reviews
4.3
3 ratingsSample
- Aiko N.Sample
Good on reasoning models and when they are worth it. Would like a template for the comparison sheet, but I built one from the lessons.
- Olga T.Sample
As a product manager I needed a decision process, not a list of names. The cost per successful answer idea changed our comparison completely.
- Diego M.Sample
We ran the comparison on 80 real tickets and a smaller model won on cost with equal accuracy. Practical and refreshingly free of hype.
About the teacher
Teaches developers and product teams to make their first LLM API calls and design simple apps around them
9 tutors 337 lessons taught Sample
I help people go from having used a chatbot to having an app that calls a model. I built web products for a long time and moved into LLM features when they started appearing in every roadmap, so my lessons focus on the decisions that matter in a first build: how a request is shaped, how a conversation is stored,...
See Gabriela's profile and tutorsMore like this
Other tutors on the same or nearby topics.