Models, limits & cost

The single most useful habit on the free tier: knowing which AI model to reach for, and when. Most beginners burn through their daily limit on the strongest model doing work a cheaper one would have done faster — and then can't work for hours. Here's how to avoid that.

The three models, in plain words

You'll meet a small family of models. They trade depth against speed and cost — the deeper one "thinks" harder but is slower and uses up more of your allowance.

The rule: volume → cheap, leverage → strong

Here is the whole policy in one line:

Volume → a cheap model. Leverage → a strong one.

• High-volume, routine work (formatting a page, adding fields to a form, repetitive edits) → Sonnet. It's faster and won't eat your allowance.
• Low-volume, high-leverage work (deciding what to build, reviewing a plan, fixing something that has you stuck) → Opus. This is where the deep model earns its cost.
• Don't want to think about it? Run opusplan — an automatic hybrid where the strong model plans and the fast one builds. You get the judgment where it matters and the speed everywhere else.

Switching mid-session

You don't pick a model once and live with it. Switch whenever the work changes character — type the command right in the chat:

/model sonnet     # routine building — fast and cheap
/model opus       # stuck on something hard — bring in the deep thinker
/model opusplan   # let it auto-switch: plan strong, build fastCopy

A natural rhythm: stay on sonnet for the steady work, flip to opus the moment you hit something that genuinely needs untangling, then flip back. Or just run opusplan and let it decide for you.

Managing your limits without frustration

On the free tier you have a daily allowance. The way to never feel cornered by it:

• Default to the cheaper model. Most of what you do is routine — let it be cheap. Save the strong model for the few moments that need it.
• Front-load the thinking. Use the strong model to make the plan, then let the cheap one execute it. One expensive decision beats ten expensive attempts.
• Notice when you're idling. If you find yourself waiting hours for the limit to reset, that's the signal worth sitting with: a paid tier that lets you keep shipping — and earning — usually beats idling for the reset. We say this honestly, not to push you: the whole point of the Academy is that the always-free craft is genuinely enough to build real things. But once your ideas are making or saving you money, paying for the tool that makes them is no longer a cost — it's leverage. (More on that line in what's always free, and what's not.)

A note that won't go out of date

Where next