Calculate Claude API cost from token counts
Select a Claude model, then enter the input and output tokens for one request. If you know the expected number of requests per month, enter it to project monthly spend. The prices shown are standard Anthropic API base rates for uncached text requests. Check the official Anthropic pricing for rates and thresholds that apply to your workload.
For example, with Claude Sonnet 5.5 at $2 input and $10 output per million, 5,000 input and 1,000 output tokens cost $0.02. At 10,000 such calls per month, the estimate is $200. If your average reply is 500 tokens, the same volume would cost $150 at those rates.
Get the Claude input-token count
For an actual Anthropic request, use its token counting endpoint with the planned messages, system instructions, and tools. Enter that count directly here. The optional paste mode uses a general OpenAI reference encoding, so its result is a rough proxy for Claude. After a representative call, use the response's usage fields to refine both input and output assumptions.
Factor in caching and longer requests
A cache read, a cache write, and regular input can each have a different rate. Price each category separately using Anthropic's usage breakdown. Claude Haiku 5.5 has a higher rate for prompts over 100,000 input tokens. The default Haiku preset uses the lower tier. For an eligible long prompt, use the current higher rate through Custom rates.
Use the cost-per-token converter when comparing unit rates across providers, or price a full month's Claude usage in the token cost calculator.