Google has given a small number of companies access to an early version of its Gemini generative artificial intelligence model.
The Information reports that the company is close to incorporating it into its consumer services and selling it to enterprises through Google Cloud. Gemini is intended to compete with OpenAI’s GPT-4, which is available through Microsoft Azure.
Gemini is made up of multiple large-language models able to write text, images, and code. Google employees told The Information that the models should produce fewer hallucinations than GPT-4, as it was also trained on Google’s proprietary data from its consumer products.
The company plans to make Gemini available through its Google Cloud Vertex AI service, and will offer access to models of different sizes so that developers can use smaller ones for simpler tasks. It has given developers access to a relatively large version of Gemini but not the largest version it is working on.
In May, Google said that it would offer Palm 2 LLMs to Google Cloud customers through Vertex.