MAI-1 from Microsoft Competes with OpenAI and Develops a New LLM Model

MAI-1 is the new addition to Microsoft’s artificial intelligence list of projects.

May 9, 2024 - 11:34
 0
MAI-1 from Microsoft Competes with OpenAI and Develops a New LLM Model

Microsoft is developing a new large language model named MAI-1 and competes with OpenAI, Google and Anthropic. According to a report by The Information, the project is overseen by Mustafa Suleyman, Microsoft’s AI CEO. The purpose of the new LLM model is not clear yet but it will be larger than previously released open-source LLM models by Microsoft. 

MAI-1, GPT4 or Grok?

MAI-1 is the first in-house LLM of this magnitude. It will have 500 billion parameters, making it one of the largest language models developed by Microsoft. MAI-1 surpasses all open-source models developed by the tech giant, including the recently released Phi-3 mini

GPT4 is rumored to have 1.74 trillion parameters according to its running speed. Grok-1 from Elon Musk has 314 billion parameters and is considered the biggest open-source LLM. DeepMind and Claude 3 did not disclose the number of their AI model parameters. 

Why build MAI-1?

It’s not clear why Microsoft is developing a new LLM model while investing $10 billion in OpenAI. According to The Information, the purpose of MAI-1 will depend on its performance. The tech giant will train the model using various data sources and will use Nvidia GPUs to build a cluster of servers. 

Kevin Scott, the CTO of Microsoft said on LinkedIn that he was not sure why this was news; however, he confirmed the development of MAI-1. 

Source: LinkedIn

Microsoft is taking the lead in the AI race and has integrated OpenAI’s technology across its products including Copilot and Dynamics 365 AI. The Information reported that Microsoft may preview MAI-1 as early as its Build developer conference later this year. Previewing the model depends on the progress made in the coming weeks.  

Microsoft is going all-in on AI

Microsoft is heavily involved in artificial intelligence and large language models. Last month, it launched a lightweight AI model called Phi-3 mini. According to Sébastien Bubeck, Microsoft’s vice president of GenAI research, the Phi-3 mini is dramatically cheaper and it’s ten times more cost-effective than other models with similar capabilities.  

Today, Microsoft announced a new AI data center in Wisconsin. President Joe Biden joined the tech giant to make the announcement. Microsoft is continuously expanding its partnerships in the AI field. Two months ago, Microsoft partnered with the French company Mistral AI. It has also partnered with Vodafone for ten years to integrate generative AI into the telecom company. 

What's Your Reaction?

like

dislike

love

funny

angry

sad

wow

CryptoFortress Disclosure: This article does not represent investment advice. The content and materials featured on this page are for educational purposes only.