Glossary

AI API terms in plain language

Short definitions for the terms used across the model catalog, documentation, billing pages, and usage logs.

Last updated: 2026-08-28

Base URL and API key

The Base URL tells an SDK where to send requests. An API key authenticates the account and must be kept secret. HippoAPI uses one Base URL and allows keys to be scoped through console controls.

Token, context window, and rate limit

A token is a unit used by many text models for input and output. The context window is the maximum combined content a model can consider. A rate limit controls how quickly or concurrently requests may be sent.

Modality and compatible endpoint

A modality is a kind of input or output such as text, image, audio, or video. An OpenAI-compatible endpoint follows a familiar request shape, but each model can still support a different subset of parameters and behavior.

Input, output, cache, and per-request pricing

Input pricing applies to content sent to a model; output pricing applies to generated content. Cache rates can apply to reused prompt content. Some media or specialized models charge per request or another published unit instead of tokens.