What is the difference in pricing when choosing model strength: low, high, max, xhigh? I’m talking about Fable
Good question. The strength level (low, medium, high, xhigh, max) does not change Fable 5’s per-token price. It only changes how much the model “thinks,” which in turn affects how many tokens a request uses.
As an example, you could have Fable on Max and you could have a prompt that is Say HELLO and only HELLO and it likely won’t use that many more tokens than it would if it was on low, because you’ve given such as basic task. But if you give it a more open ended task, for example "Research and find me the best car for my budget - and you select Max - it’s going to spend a lot more tokens doing that than if you had it on low. On low it’s going to get you answer faster and with fewer tokens while Max is going to spend a lot more time thinking to give you the best answer.
Fable 5 is billed at $10 per 1M input tokens and $50 per 1M output tokens, no matter which level you pick. Reasoning/thinking tokens count as output, so higher levels generate more tokens and cost more per request, even though the rate never changes.
Rule of thumb:
- low/medium: cheaper, for simpler tasks
- high: the default, best for most work
- xhigh/max: much larger thinking budgets for hard, long-running tasks (can cost several times more per request)
So the difference you see isn’t a price multiplier, just more tokens used.
Docs: Claude Fable 5 | Cursor Docs