How AI providers charge for access to their models. The dominant model is per-token pricing — you pay for the number of tokens you send (input) and receive (output), with output tokens typically costing 3-5x more. Other models include per-request pricing, monthly subscriptions, committed-use discounts, and free tiers. The race to lower prices has been fierce, with costs dropping 10-100x in two years.
Why it matters
Pricing determines what you can build. An application that makes 10,000 API calls per day lives or dies by the per-token cost. Understanding pricing models, comparing providers, and optimizing token usage is a core skill for anyone building AI-powered products.