The 2023-2026 model price war — and why your cost model is already obsolete — step 7 of 8
Write cost_per_user_month(usage) that takes a usage profile dict
and returns the projected dollar cost per user per month as a float
rounded to 2 decimals.
usage has four fields:
messages_per_day: int — how many model calls per day for this useravg_input_tokens: int — average input tokens per callavg_output_tokens: int — average output tokens per callmodel: str — one of"haiku","sonnet","opus"
Use these 2026-real prices per million tokens (input / output):
haiku: $1.00 / $5.00sonnet: $3.00 / $15.00opus: $5.00 / $25.00
Treat one month as 30 days.
Two demo users run. Expected output:
power user (haiku): $1.72/user/mo
pro user (sonnet): $4.14/user/mo
⌘↵ runs the editor.read, then continue.
Write cost_per_user_month(usage) that takes a usage profile dict
and returns the projected dollar cost per user per month as a float
rounded to 2 decimals.
usage has four fields:
messages_per_day: int — how many model calls per day for this useravg_input_tokens: int — average input tokens per callavg_output_tokens: int — average output tokens per callmodel: str — one of"haiku","sonnet","opus"
Use these 2026-real prices per million tokens (input / output):
haiku: $1.00 / $5.00sonnet: $3.00 / $15.00opus: $5.00 / $25.00
Treat one month as 30 days.
Two demo users run. Expected output:
power user (haiku): $1.72/user/mo
pro user (sonnet): $4.14/user/mo
this step needs the editor
on desktop today; in the app (coming soon). save your spot and we'll bring you back here when you're ready.