When an app builder asks you for an API key, or someone says a tool “uses another model,” they’re talking about what happens behind the scenes when a program uses a model. That’s true of all three ways to use a model from Request, Chat, and Agent, and some tools and skills bring a second model into the job.
APIs and API Keys
An API is how one program asks another service for something. When an app sends a request to the company that runs the model and gets an answer back, developers say the app “calls the API.” Every message you send in a chat app reaches the model this way. The chat app uses the company’s own access, so you never see a key.
An API key is like a pass that proves a program is allowed to use a service. It also ties every request to an account, and requests made with a key are usually charged by use, which works differently from a chat subscription. That’s the pay-per-use side of Usage, Limits, and Tokens. App builders ask for keys when your app needs a service of its own, like a model, email, or payments.
Treat a key like a password. Paste it only into the settings box the app builder made for keys. In an ordinary chat message, it becomes part of the conversation, and the whole conversation gets sent to the model.
