Model selection
Picking the right model for the job
Most AI apps let you choose between several models. They are not all the same. Picking the right one makes your work faster, cheaper, and better.
Why there are different models
Think of vehicles. A bicycle is quick and cheap for a short trip. A truck is slow and costly, but it can move a whole house. You pick based on the job.
AI models are similar. Each company offers a family of models:
- Some are big and powerful.
- Some are small and fast.
- Some have extra skills, like thinking longer or reading images.
In chat apps, you'll find these in the model picker, usually a menu near the top of the chat or next to the message box.
Big vs small models
Big models are the most capable. They handle hard tasks, long documents, and tricky code. They are slower and cost more.
Small or fast models are quick and cheap. They are great for simple tasks like short emails, summaries, or sorting text.
Reasoning or thinking modes
Some models can "think" before they answer. They work through the problem step by step first, like doing math on scrap paper.
- Good for: planning an app, debugging, math, and multi-step logic.
- Trade-off: answers take longer and use more tokens.
Look for a toggle or option with a name like "thinking", "reasoning", or "extended thinking".
Other things to compare
- Multimodal: a multimodal model can work with more than text, like images, PDFs, or audio. Useful if you want to share a screenshot of a bug.
- Context size: how many tokens fit in its context window. Bigger is better for long documents.
- Cost: in chat apps, it affects your usage limits. With an API, you pay per token. Check the pricing page.
| Feature | Good choice | Why |
|---|---|---|
| Quick email or summary | Small, fast model | Cheap and instant |
| Planning or building an app | Big model | Handles complex instructions |
| Tricky bug or logic puzzle | Reasoning / thinking mode | Works step by step |
| Screenshot or PDF | Multimodal model | Can read images and files |
| Very long document | Large context window | Fits more text at once |
| AI feature in your app | Smallest model that works well | Keeps API costs low |
Try it yourself
Send the same prompt to two different models in the model picker and compare:
I want to build a simple habit tracker web app.
List the 5 most important features and explain each in
one sentence. Then point out one thing beginners often forget.Which answer was faster? Which was more useful? This simple test is the best way to choose.
Common mistakes
- Always using the biggest model. For simple tasks, it's slower and costs more for no real gain.
- Always using the fastest model. For planning and building apps, a weak model can waste hours.
- Choosing by name or hype. Test with your real task instead.
- Forgetting thinking mode exists. If an answer to a hard problem is sloppy, turn it on and ask again.
Practice
Try these exercises to lock in what you just learned.
What's next?
You now know the key ideas behind AI and LLMs. Time to put them to work. Next, let's meet the AI chat tools you'll use most: ChatGPT, Claude, and Gemini.