Microsoft is preparing a new large-scale artificial intelligence (AI) model, with enough capacity to compete against similar models from Google and OpenAI, the company responsible for the ChatGPT platform.
The information was reported by The Information newspaper and quoted by Reuters.
The development of the new model – which is known internally as MAI-1, and whose purpose is not yet fully known – is being overseen by Mustafa Suleyman, co-founder of Google DeepMind and former CEO of AI startup Inflection.
According to the sources heard by Information, the MAI-1 will be “much bigger” than the previous, smaller, open-source models previously trained by Microsoft. It will, it is estimated, be more expensive than previous models.
The MAI-1 will have around 500 billion parameters at its disposal, while OpenAI’s GPT-4 will have a billion parameters.
According to Reuters, Microsoft launched a smaller artificial intelligence model last month called Phi-3-mini. The aim is to attract a wider customer base with affordable options.
Leave a Reply