Skip to content

Pick your style: Denkbaar

Book a conversation

Articles

Knowledge

What is an API?

An API is a connection that lets software use a service or a model.

A person studies a tablet connected to a server.
An API lets your application use a service. Illustration: Denkbaar.

An API is a connection that lets software use a service, for example an AI model. That way you can call a model from your own application, instead of only using it in a chat.

What you read about it

  • OpenAI announced GPT-5.5 on 23 April 2026. A day later the model was also available through the API. These were two separate events: GPT-5.5.
  • Google made Gemini 2.0 Flash generally available through the API on 5 February 2025. That connection lets software call the model: Gemini 2.0 Flash.
  • A model used through an API runs at the provider. An open model you can run on your own server: Claude and Mistral.
  • If your system has an API or an export function, it is faster and more reliable than clicking on a screen: Claude computer use.
A chip on a stand sits beside an unplugged connector.
An announcement does not prove an API is available. Illustration: Denkbaar.

Why it matters to you

A product announcement does not prove that the connection you want is already possible. Ask what is available through the API and which model version is called. Write down what input the model gets and what answer your application expects.

A hand compares input and output beside a small server.
Define the input and response you expect. Illustration: Denkbaar.

Written on 13 November 2026, based on the articles and sources below.

Sources

  1. GPT-5.5 ("Spud"); API on 2026-04-24
  2. Gemini 2.0 Flash GA, 2.0 Pro experimental
  3. Anthropic: Computer use, Claude 3.5 Sonnet, and Claude 3.5 Haiku

Topics: apibasics