Claude model family
Pricing, capabilities & effort breakdown · source: platform.claude.com · July 2026
What does the same task cost?
The task: write 1,000 lines of Python code. One bar per model — each colored segment ends at the total cost of that effort level. Cheapest at the top.
| Claude Sonnet 5 |
$0.26
$0.61
$0.83
|
| Claude Opus 4.8 |
$0.44
$1.01
$1.39
|
| Claude Fable 5 |
$0.87
$1.27
$2.02
$2.77
|
Approximate figures at standard list prices. Higher effort levels spend more tokens on reasoning while producing the same 1,000 lines of code, so the bar grows segment by segment: the number inside each segment is the total cost of the task at that effort level. Claude Haiku 4.5 is not shown — it has no effort parameter, so there is nothing to break down. For scale: Claude subscription plans come with a usage limit that resets every 5 hours. Very roughly, one such 5-hour window is worth about $5–10 of model usage on Pro and around $25–100 on Max plans — so on Pro, a window buys you roughly a dozen tasks like this on Sonnet 5 at High effort, or a handful on Opus 4.8.