What does each Claude model cost per million tokens?
Per Anthropic's API pricing page, all prices are in USD per million tokens (MTok).
| Model | Input | Output | 5-minute cache write | 1-hour cache write | Cache hit |
|---|---|---|---|---|---|
| Claude Haiku 5.5 (prompts up to 100,000 tokens) | $0.10 | $0.50 | $0.125 | $0.20 | $0.01 |
| Claude Haiku 5.5 (prompts over 100,000 tokens) | $0.50 | $2.50 | $0.625 | $1 | $0.05 |
| Claude Sonnet 5.5 | $2 | $10 | $2.50 | $4 | $0.10 |
| Claude Opus 5.5 | $4 | $20 | $5 | $8 | $0.20 |
| Claude Fable 5.2 | $8 | $40 | $10 | $16 | $0.20 |
As reviewed on November 19, 2026. Anthropic also lists earlier models at their own rates; see its page. Prices can change, so check the page before you budget.
How do batch, caching and tools change the price?
Each is a separate line on Anthropic's pricing page.
- Batch API: a 50 percent discount on both input and output tokens for asynchronous processing of large volumes of requests.
- Prompt caching: reading from cache is billed at a fraction of the input price (the table shows each model's cache hit rate), and Anthropic says these multipliers stack with the batch discount.
- Web search: $10 per 1,000 searches, plus standard token costs for the content it returns. Web fetch has no additional charge beyond tokens.
- Cloud marketplaces: Anthropic's page also lists Claude on other cloud platforms, including Amazon Web Services, billed through the cloud marketplace in Claude Consumption Units at $0.01 USD per CCU.
- US-only inference: for Claude 4.6 and later models, specifying US-only inference applies a 1.1x multiplier on all token prices.
How do you get a Claude API key and cap your spend?
Per Anthropic's get started guide, you create a Claude Console account, then create an API key in the Console.
If you want a developer to build with the API, see our AI implementation page.
- Steps: sign up or sign in to the Claude Console, create an API key, then set it as an environment variable so the SDKs read it.
- Spend caps: per Anthropic's rate limits page, each of the Start, Build and Scale usage tiers has a monthly spend cap of $500, $1,000 and $200,000, and you can set your own lower limit in the Console. At the cap, requests return HTTP 429 until the next month.
- Subscriptions are separate: a Claude Pro or Max plan does not appear on the API pricing page; the API is billed per token. Max and Ultra list monthly API credits, per the plans guide.
Frequently asked questions
How much does the Claude API cost?
Anthropic lists base rates per million input and output tokens of $0.10 and $0.50 for Haiku 5.5 (prompts up to 100,000 tokens), $2 and $10 for Sonnet 5.5, $4 and $20 for Opus 5.5, and $8 and $40 for Fable 5.2.
How do I get a Claude API key?
Create a Claude Console account, then create an API key in the Console and set it as an environment variable, per Anthropic's get started guide.
Is there a discount for batch processing?
Yes. Anthropic's Batch API gives a 50 percent discount on input and output tokens for asynchronous processing.
Does prompt caching save money?
Anthropic lists cache hits at a fraction of the input price, for example $0.10 per million tokens on Sonnet 5.5 against a $2 input rate.
Can you build with the Claude API for us?
Yes. We scope what to build, set spend limits and keep keys safe. Book a call to start.
Need help with Claude?
We help small and mid-size teams choose, set up and run AI tools. See our Claude guides and services, and the Claude enablement hub, and the AI implementation.



