A worked estimate you can check line by line. Change the numbers to your own in the cost calculator.
What we assume
A 1,500-token system prompt, about 6,000 tokens of retrieved document chunks plus the question, and a 500-token answer with citations.
Cost per question
| Model | No caching | With caching | Per 1,000 questions, with caching |
|---|---|---|---|
| Claude Fable 5.2 | $0.080 | $0.068 | $68.30 |
| Claude Opus 5.5 | $0.040 | $0.034 | $34.30 |
| Claude Sonnet 5.5 | $0.020 | $0.017 | $17.15 |
| Claude Haiku 5.5 | $0.0010 | $0.00086 | $0.865 |
Cost per month
| Model | One user (200 questions/month) | 1,000 users |
|---|---|---|
| Fable 5.2 | $13.66 | $13,660 |
| Opus 5.5 | $6.86 | $6,860 |
| Sonnet 5.5 | $3.43 | $3,430 |
| Haiku 5.5 | $0.173 | $173 |
Where the money goes
- Cost scales with how much you retrieve. Going from 6,000 to 3,000 retrieved tokens roughly halves the input cost.
- Retrieved chunks change with every question, so they are not cached; only the fixed system prompt is.
- If many users ask about the same document, put that document in the cached prefix instead of retrieving pieces of it.
To check what a plan price leaves you after these costs, use the AI SaaS margin calculator. To see how much caching saves on your own prompt sizes, use the prompt caching savings calculator.
Questions
How much does one question cost with the Claude API?
With prompt caching, from about $0.00086 on Claude Haiku 5.5 to about $0.068 on Claude Fable 5.2, for a question of 1,500 reused + 6,000 new input tokens, 500 output tokens. Without caching it is $0.0010 to $0.080.
How much does it cost per month?
One user (200 questions/month): about $0.173 on Haiku 5.5 up to $13.66 on Fable 5.2. 1,000 users: about $173 on Haiku 5.5 up to $13,660 on Fable 5.2, with caching.
How can I make it cheaper?
Cost scales with how much you retrieve. Going from 6,000 to 3,000 retrieved tokens roughly halves the input cost.