Microsoft is reportedly readying its own AI model called MAI-1

The LLM reportedly will have approximately 500 billion parameters

Reading time icon 3 min. read


Readers help support Windows Report. We may get a commission if you buy through our links. Tooltip Icon

Read our disclosure page to find out how can you help Windows Report sustain the editorial team Read more

Microsoft is reportedly readying its own AI model called 'MAI-1'

Microsoft is reportedly developing a new AI model, internally called MAI-1. The world is going crazy behind AI and that’s fair as it is the booming tech. Microsoft is among some of the big global tech companies that have taken a keen interest in AI.

The company is rapidly putting out AI products by leveraging OpenAI LLMs like GPT-4. Talking of large language models, Microsoft doesn’t have an impactful one that can shake its market rivals. But, that will change soon.

MAI-1 is under development and being overseen by ex-Google AI leader

Now it seems Microsoft is readying a new AI model to compete against rival companies like Google, Anthropic, and OpenAI. The news comes from The Information (via Reuters) which reports the company is developing MAI-1, the new AI model, which will take on leading LLMs in the market.

Mustafa Suleyman, who happens to be an ex-CEO of AI startup Inflection and co-founder of Google DeepMind is in charge of Microsoft’s new AI model MAI-1.

Not to forget, Microsoft recently paid Inflection $650 million for the rights to the startup’s intellectual property back in March. It even hired a majority of Inflection’s staff.

Well, that doesn’t mean Microsoft is carrying forward the AI model from Inflection. However, there are possibilities that Microsoft could use other tech and data to train its new AI model though.

‘MAI-1’ will be much larger than the previous models launched by Microsoft

As per the report, the new AI model MAI-1 will be much larger than the previous smaller models. This eventually means that it would be more expensive than all other models.

Microsoft is reportedly readying large servers using a bunch of Nvidia’s GPUs and tons of data to train and improve the new AI model.

The report also mentions that MAI-1 will have approximately 500 billion parameters less than OpenAI’s GPT-4 one trillion parameters. Not to forget, the recently announced Phi-3 mini model has 3.8 billion parameters.

There’s not much information available about Microsoft’s new AI model ‘MAI for now, but going down the line pictures will surely get clearer.

That’s not all, the recent report has also sparked a new conversation on whether Microsoft will continue to use OpenAI’s model for its products. Or, MAI-1 will be used by the company as a replacement. Who knows, time will tell.

Are you excited about MAI-1? Share your thoughts on whether or not it would be a great decision for the company going forward.

More about the topics: AI, microsoft