Glossary
API
An API (application programming interface) is the doorway programs use to talk to each other. "Using AI via API" means an app sends prompts straight to the model and pays per token, instead of a person typing in a chat window.
You can use AI through the front door (the chat app) or through the service entrance: the API. Same models behind both, different deal. The chat is a flat subscription for humans; the API is metered access for software, billed per token, like electricity.
Why would a non-programmer care? Three reasons keep coming up. First, every “AI-powered” app you use, the invoice tool that summarizes or the booking widget that chats, is almost certainly calling someone’s API behind the scenes; you’re an API user already. Second, the terms differ: major providers don’t train on API data by default, which matters when a tool handles your customers’ information. Third, for heavy, regular workloads, API pricing can beat subscriptions, or vastly exceed them, depending on volume.
Where you’ll meet this
In product fine print (“powered by GPT-5,” “built on Claude”), in developer dashboards at OpenAI, Anthropic, Google, and Mistral, and in price comparisons quoting “per million tokens.” If you’re weighing chat subscription versus API for your usage, our plan picker covers that comparison.