Google releases API for Gemini 1.5 Pro

Jeff Dean, chief scientist at Google DeepMind, recently shared in a post on X that the company has begun rolling out API support for Gemini 1.5 Pro.

Huge context window for better processing

Gemini 1.5 Pro has a massive contextual window that can handle 1 million tokens. This represents a huge jump from 128,000 tokens for OpenAI's GPT-4 and 32,000 tokens for the current Gemini Pro, and expanding the context window allows a wider range of information to be considered at once – from text to images to video to audio and code. The Gemini 1.5 Pro is said to be capable of processing and analyzing one hour of video, 11 hours of audio, and codes containing more than 30,000 lines or more than 700,000 words.

Gemini 1.5 Pro is based on the so-called Mixture of Experts (MoE) architecture, which consists of a variety of smaller, specialized language models. Each of these models is aimed at specific areas of responsibility. This differentiated approach increases the overall efficiency of the system by allowing it to analyze and interpret large amounts of data simultaneously – whether it is text, images, audio, video or code. To use the API, developers must have a Gemini Pro account and request access to the Gemini API in Google MakerSuite.

Google Gemini coming soon to iPhone?