How AI API token pricing works

API providers usually quote separate prices for input and output tokens. For a rate quoted per million tokens, multiply each token count by its rate and divide by one million, then add the two amounts. Reasoning tokens may be billed as output; cached input and other provider-specific charges can change the bill. Benchmark-run cost measures a complete workload instead of a fixed token quantity, so it captures differences in token usage between models.

Which should you choose?

No single benchmark covers every use case. Try your own prompts and compare accuracy, response time, input/output support, and the actual bill.

Estimated token cost

ModelInput tokensOutput tokens

Agnes 2.5 Pro Beta: $0.40

Input price / 1M tokens (USD): 0.10
Output price / 1M tokens (USD): 0.30

This estimate uses the listed input and output rates. Caching, tools, reasoning tokens, long-context tiers, taxes and hosting can change the bill.

Guides

Best-value AI models Cheapest AI APIs: compare token prices Fastest AI models: compare output speed Open-weight AI models: benchmarks and pricing How AI API token pricing works

How we compare models

Intelligence uses the Artificial Analysis Intelligence Index. Benchmark cost is the total USD cost of running that test, including reasoning tokens; it is not a per-token API price. Rankings use the same dataset and calculations in every language. General benchmark scores do not measure performance in a particular language or task.