Claude API pricing questions
How much does the Claude API cost?
It depends on the model. Per million tokens, Claude Haiku 5.5 costs $0.10 input and $0.50 output, Sonnet 5.5 $2.00 input and $10.00 output, Opus 5.5 $4.00 input and $20.00 output, and Fable 5.1 $10.00 input and $50.00 output. A chat reply of 3,000 input and 500 output tokens costs $0.0006 on Haiku 5.5 and $0.011 on Sonnet 5.5.
What is the cheapest Claude model?
Claude Haiku 5.5, at $0.10 input and $0.50 output per million tokens for prompts up to 100,000 tokens. Longer prompts cost $0.50 input and $2.50 output. With the Batch API it drops to $0.05 input.
What is the most expensive Claude model?
Claude Fable 5.1 and Claude Mythos 5.1, at $10.00 input and $50.00 output per million tokens. Mythos 5.1 has limited availability. Their cache reads cost $0.25 per million, the lowest share of the input price of any Claude model.
Is there a free tier for the Claude API?
There is no ongoing free tier. New accounts get a small amount of free credit for testing, and after that every token is billed at the prices on this page. Anthropic sales can arrange extended trials for enterprise evaluations.
Do Batch and prompt caching discounts stack?
Yes. The Batch API halves every price, including cache reads, so a cached input token on Sonnet 5.5 in a batch costs $0.05 per million instead of $2.00.
Why did my Claude bill go up after switching models?
Claude 4.7 and later models use a newer tokenizer that counts about 30% more tokens for the same text. A model with a lower price per token can still cost more per request, so compare the cost of real requests, not only the price list.
Which Claude models are being retired?
Claude Sonnet 4.5 retires on the Claude API on November 30, 2026; Anthropic recommends Sonnet 5.5 instead. Opus 4, Opus 4.1 and Sonnet 4 have already shut down on the Claude API, though some cloud platforms still serve them.
Is Claude cheaper on AWS Bedrock or Google Cloud?
Amazon Bedrock and Google Cloud set and publish their own Claude prices and bill you directly, so check their price lists. For Claude 4.5 and later models, their regional and multi-region endpoints cost 10% more than global endpoints. On the Claude API, keeping inference in the US costs 1.1×.
Does the Claude API charge more for long prompts?
Not on Claude 4.6 and later models: the 1M-token context window is billed at standard prices. The exception is Haiku 5.5, which charges $0.50 input and $2.50 output per million tokens once a prompt passes 100,000 tokens.
How do I track Claude API cost per customer?
Tag each Claude call with the customer it served, then price it at these rates and add it up by customer. Ganivra does this automatically with two lines of the Python SDK, and shows each customer's Claude cost next to what they pay you.
More detail: Claude Opus pricing · Claude Sonnet pricing · OpenAI API pricing · Gemini API pricing · all supported models and prices.