What every current Claude model costs through the API, what a typical request costs in practice, and where caching and batching cut the bill.
Prices per million tokens
| Model | Input | Output | Cache write 5m | Cache write 1h | Cache read | Batch in / out |
|---|---|---|---|---|---|---|
| Claude Fable 5.2 | $8 | $40 | $10 | $16 | $0.2 | $4 / $20 |
| Claude Opus 5.5 | $4 | $20 | $5 | $8 | $0.2 | $2 / $10 |
| Claude Sonnet 5.5 | $2 | $10 | $2.5 | $4 | $0.1 | $1 / $5 |
| Claude Haiku 5.5 | $0.1 | $0.5 | $0.125 | $0.2 | $0.01 | $0.05 / $0.25 |
- Claude Haiku 5.5: prompts over 100,000 tokens are billed at $0.5 input and $2.5 output per million tokens for the whole request.
What a single request costs
Without caching or batching:
| Request | Fable 5.2 | Opus 5.5 | Sonnet 5.5 | Haiku 5.5 |
|---|---|---|---|---|
| Short chat reply 2,000 in / 300 out | $0.028 | $0.014 | $0.0070 | $0.00035 |
| Agent turn with tools and history 12,000 in / 800 out | $0.128 | $0.064 | $0.032 | $0.0016 |
| Summarize a 40-page document 30,000 in / 1,000 out | $0.280 | $0.140 | $0.070 | $0.0035 |
Model vs model
- Claude Fable 5.2 vs Opus 5.5
- Claude Fable 5.2 vs Sonnet 5.5
- Claude Fable 5.2 vs Haiku 5.5
- Claude Opus 5.5 vs Sonnet 5.5
- Claude Opus 5.5 vs Haiku 5.5
- Claude Sonnet 5.5 vs Haiku 5.5
What a typical app costs
- How much does a Claude chatbot cost to run?
- How much does a Claude customer support bot cost?
- How much does a Claude coding agent cost per task?
- How much does a Claude RAG app cost per question?
- How much does it cost to summarize documents with Claude?
- How much does classification or data extraction with Claude cost?
Questions
How is Claude API usage billed?
Per token. Input tokens (your prompt, conversation history, tool definitions and documents) and output tokens (the reply, including extended thinking) have separate prices per million tokens. There is no fixed monthly fee for the API itself.
What is the cheapest Claude model?
Claude Haiku 5.5 at $0.1 input / $0.5 output per million tokens.
How much does prompt caching save?
A cache read costs a small fraction of the normal input price, so a long system prompt, tool list or document that repeats across requests becomes much cheaper after the first call. Writing to the cache costs a little more than normal input (more for the 1-hour cache than the 5-minute one).
What does the Batch API cost?
Requests sent through the Message Batches API are billed at 50% of the normal price, in exchange for results arriving asynchronously (usually within 24 hours).