Microsoft has developed an AI model that can compete with OpenAI.
Microsoft is reportedly developing its own large language model (LLM) for artificial intelligence (AI), competing with products from OpenAI, Google, or Meta. The latest LLM by Microsoft is called MAI-1, which contains 500 billion parameters, making it the largest LLM the company has ever developed. However, it is smaller than OpenAI GPT-4's one trillion parameters.
Previously, Microsoft had been using products from OpenAI, which the company has invested over 10 billion USD in for many years. MAI-1 is the first LLM developed by Microsoft, following the announcement of the small-scale model Phi-3 mini with 3.8 billion parameters, serving a number of services such as Azure, Hugging Face, and Ollama.
MAI-1 requires a lot of computing power and training data. Microsoft has reportedly reserved a large number of servers equipped with Nvidia chips to serve this project. The project is overseen by Mustafa Suleyman, co-founder of Google DeepMind and former CEO of AI startup Inflection. In March, Microsoft spent $650 million to attract most of the personnel and models developed by AI Inflection. However, sources say that MAI-1 is an independent product of Microsoft, not transferred from Suleyman's old company. The company has not yet commented.
Microsoft is a leader in AI among BigTech companies, and while it does not own a famous LLM, it has contributed greatly to promoting ChatGPT and OpenAI through investment deals and its relationship with CEO Sam Altman. Microsoft's Copilot chatbot running on GPT-4 Turbo is also considered one of the best chatbots today. The company may introduce MAI-1 to developers at the Build event, scheduled to be held at the end of May.
Previously, Microsoft had been using products from OpenAI, which the company has invested over 10 billion USD in for many years. MAI-1 is the first LLM developed by Microsoft, following the announcement of the small-scale model Phi-3 mini with 3.8 billion parameters, serving a number of services such as Azure, Hugging Face, and Ollama.
MAI-1 requires a lot of computing power and training data. Microsoft has reportedly reserved a large number of servers equipped with Nvidia chips to serve this project. The project is overseen by Mustafa Suleyman, co-founder of Google DeepMind and former CEO of AI startup Inflection. In March, Microsoft spent $650 million to attract most of the personnel and models developed by AI Inflection. However, sources say that MAI-1 is an independent product of Microsoft, not transferred from Suleyman's old company. The company has not yet commented.
Microsoft is a leader in AI among BigTech companies, and while it does not own a famous LLM, it has contributed greatly to promoting ChatGPT and OpenAI through investment deals and its relationship with CEO Sam Altman. Microsoft's Copilot chatbot running on GPT-4 Turbo is also considered one of the best chatbots today. The company may introduce MAI-1 to developers at the Build event, scheduled to be held at the end of May.
Share