Model or Effort? The Question That Trips Everyone Up

In short

In front of strange behavior, one question only: did it not understand, or did it not try?
It does too much (widens the job, touches unrelated things) → almost always too much effort.
It plays lazy (cuts corners, says it can't) → almost always too little effort.
Switching models is the most expensive move: try it after, not first.

The wrong reflex

When Claude Code does something strange, everyone's first thought is always the same: "I need the more powerful model." And it's almost always the wrong thought — the one you pay dearly for without fixing anything.

The reason is that most strange behavior doesn't come from how much the model knows, but from how much effort it put in. Two symptoms make it obvious.

Symptom 1 — it does too much

You ask for something small and end up with a construction site: it touched files that had nothing to do with it, "fixed" things you never asked about, rewrote parts that were already fine.

It didn't misunderstand: it understood, and then it tried too hard. With effort maxed out, a tiny task becomes an opportunity to think big. The dial to turn is effort, downward — not the model.

Symptom 2 — it plays lazy

The opposite: you ask for something you know perfectly well is within its reach, and it tells you it can't, or does the bare minimum and stops there.

Here too, before assuming the model is too small: the effort is set low. Low effort looks for the shortest way to be done, and sometimes the shortest way is giving up. Raise the effort and try again: it costs far less than moving up a model.

When it really is the model

The model is the culprit when the problem isn't effort but understanding: it keeps proposing solutions that don't hold up, it doesn't see a contradiction that's obvious to you, it circles the same mistake even after you've given it more to go on.

At that point you're not asking for more effort, you're asking for more judgment: move up a tier.

And now

There's still the money question. In the next lesson we close the chapter with the part nobody tells you: why choosing the cheap model can cost you more, what actually gets consumed during a session, and why you don't switch engines halfway through.

Did you get it?

Two minutes, no grade. Trying to answer — even getting it wrong — cements the concepts far more than rereading. Not in the mood? Skip it: the lesson still counts as complete.

You ask it to change a button's color and it rewrites half the project. What do you turn?

It tells you it can't do something you know it can, or it cuts corners. What do you turn?

What's the question to ask yourself in front of strange behavior?

1
1. 1
2. 2
3. 3
2
1. 4
2. 5
3. 6
4. 7
3
1. 8
2. 9
3. 10
4. 11
5. 12
4
1. 13
2. 14
3. 15
4. 16
5. 17
6. 18
5
1. 19
2. 20
3. 21
4. 22
5. 23
6. 24
6
1. 25
2. 26
3. 27
4. 28
5. 29
6. 30
7
1. 31
2. 32
3. 33
4. 34
8
1. 35
2. 36
3. 37
4. 38
9
1. 39
2. 40
3. 41
10
1. 42
2. 43
3. 44
11
1. 45
2. 46