GPT-4, the much anticipated new large language AI model from ChatGPT-maker OpenAI, could be released next week according to Microsoft Germany’s CTO. The model is likely to be exponentially more powerful than its predecessor, opening up new enterprise use cases for generative AI.

Microsoft has been a partner of OpenAI since 2019 when it invested $1bn in the company, and it upped its stake by several billion dollars in the AI lab in January, following the success of ChatGPT, the AI-powered chatbot which has taken the internet by storm in recent months.

Rumours about GPT-4 and its capabilities have been swirling for some time, though OpenAI CEO Sam Altman attempted to play down what the system would be able to do in an interview earlier this year, when he said it was likely to “leave people disappointed”.

GPT-4 to be released next week?

Speaking at an event in Germany, Andreas Braun, CTO at Microsoft Germany, said the system’s arrival was imminent.

“We will introduce GPT-4 next week, there we will have multi-modal models that will offer completely different possibilities – for example videos,” said Braun in comments first reported by German media outlet Heise.

A multi-modal LLM is capable of drawing information from a variety of types of sources. This could open up the possibility of it taking information from videos and images it finds on the internet. GPT-3.5, OpenAI’s current LLM, is purely text-based.

Microsoft itself has been experimenting with multi-modal AI models already, and earlier this month released details of Kosmos-1, a model which can draw on data from text and images.

Source Tech Monitor

BDST: 1443 HRS, MAR 12, 2023
MSK