# Claude model picker cheat sheet

Model versions as shown in the video: Haiku 4.5, Sonnet 4.6, Opus 4.7.
Switch models in the Claude app with the **model selector** button.

Rule of thumb: the bigger the model, the more you pay to make it think. The smaller the model, the better it is for your wallet.

## Haiku: smallest and fastest

- Answers in a couple of seconds, most cost-efficient
- Good for simple tasks, not for analysis
- Extended thinking is available but not really useful on Haiku 4.5

Use it for:
- [ ] Quick Q&A (picking the right answer from an FAQ document for a client)
- [ ] Text classification and tagging of documents
- [ ] Moderation of chat, mail or client-facing documentation
- [ ] Data extraction from documents and images (even flipped or stained). Verify what it extracts.

## Sonnet: the everyday default

- Balances speed and analysis, uses far fewer tokens than Opus
- Handles very complex instructions, several documents at once, finding correlations
- Very reliable at following instructions, strong at coding
- Uses adaptive thinking: it decides by itself how much to reason (little for an email, a lot for comparing several reports)

Use it for:
- [ ] Content creation and editing (what works on social media, analysis of your own data)
- [ ] Research with deep research
- [ ] Code generation
- [ ] Most business use cases

## Opus: the big one

- Highest reasoning capacity, most creative output
- Slow, and can overthink simple problems
- Uses adaptive thinking, like Sonnet
- Consumes a lot of usage on the $20 plan

Use it for:
- [ ] Creative and marketing tasks
- [ ] Complex technical analysis
- [ ] Heavy code work
- [ ] Tasks where you need several pieces of output

## Which default for your plan

| Plan | Default | Go down to | Go up to |
|---|---|---|---|
| $20 | Sonnet 4.6 | Haiku for high-volume searching or tagging (for example in emails) | Opus 4.7 for marketing, creative or analysis work |
| $100 | Opus 4.7 | Sonnet 4.6 if you keep hitting the rate limit | |