Calculator

Price one request at the rates in effect when it ran. The calculator runs in your browser on the same price data as the model-prices Python package, and gives the same result as model_prices.cost() to the last digit: a test compares the two on thousands of requests, including long-context tiers, peak and off-peak hours, and thinking mode.

Input tokens exclude cache reads and writes. Longer-lived cache writes are one-hour writes for Anthropic and Moonshot AI and 12-hour writes for OpenAI; other providers charge them as cache writes. A long-context tier applies when the request's whole prompt, uncached input plus cache reads and writes, exceeds the tier's size.