Microsoft is developing an in-house artificial intelligence inference model to compete with OpenAI and may sell it to developers, citing those involved in the program.
The Washington-based Redmond, a major supporter of OpenAI, has begun testing models for Xai, Meta and DeepSeek as a potential OpenAI replacement in the case of a deputy, the report said.
Microsoft has been looking to reduce its reliance on Chatgpt Maker, even as its early partnership with startups puts it in a leading position in a profitable AI competition.
Reuters reported in December that the company has been working to add internal and third-party AI models to power its flagship AI product, Microsoft 365 Copilot, to diversify current underlying technologies from OpenAI and reduce costs.
When Microsoft announced the 365 Copilot in 2023, a major selling point was that it used OpenAI’s GPT-4 model.
According to the information report, Microsoft’s AI division, led by Mustafa Suleyman, completed training on a family of models called MAI, which performed almost the same as OpenAI’s leading model and humans for publicly recognized benchmarks.
The team is also training an inference model, which uses a reasoning process through thinking chain technology that generates answers with intermediate reasoning capabilities when solving complex problems and can compete directly with OpenAI.
The story continues below
The report said Suleiman’s team was already experimenting with exchanging MAI models, which were much larger than the earlier family of Microsoft models called Phi’s model, which is the Openai model in Copilot.
The report said the company is considering using MAI models as an application programming interface later this year, which will allow external developers to weave these models into their own applications.
Microsoft and Openai did not immediately respond to Reuters’ request for comment.
©IE Online Media Services Pvt Ltd
