Investing.com -- Last fall, Mustafa Suleyman, the head of Microsoft’s in-house artificial intelligence unit, expressed dissatisfaction with OpenAI’s lack of documentation on how its latest model, o1, worked, according to a report from The Information.

Suleyman’s role at Microsoft (NASDAQ: MSFT ), which started a year ago, is twofold. He is responsible for maintaining the partnership with OpenAI, through which the two companies share AI research and development information. Additionally, he is tasked with guiding Microsoft towards AI self-sufficiency, reducing reliance on OpenAI’s technology for most of Microsoft’s AI products.

AI researchers in Suleyman’s unit believe they have made significant progress towards the second goal, the report noted. A team led by Suleyman’s deputy, Karén Simonyan, has completed the training of a new set of Microsoft models, referred to as MAI internally. These models have reportedly performed nearly as well as leading models from OpenAI and Anthropic on commonly accepted benchmarks. The team is also training reasoning models that could compete directly with OpenAI’s models, using chain-of-thought techniques to think through problems before solving them.

Microsoft is considering releasing the MAI models later this year as an application programming interface, which will allow outside developers to integrate the Microsoft models into their own apps. This move would bring Microsoft’s models into direct competition with similar offerings from OpenAI and other AI labs.

Under Suleyman’s guidance, Microsoft has also been testing models from OpenAI’s competitors to power Copilot, a suite of AI tools built into Windows, the Edge web browser, and other Microsoft products that currently run on OpenAI’s technology. These tests include models from Anthropic and Musk’s xAI, as well as open-source models from DeepSeek and Meta Platforms (NASDAQ: META ).

Suleyman stated that Microsoft is now using models from major labs, including all open-source models, for experimentation. However, the success of his efforts to achieve more self-sufficiency for Microsoft in AI remains uncertain. While Simonyan’s team has celebrated the performance of the MAI models internally, they have not yet been released publicly, nor has Microsoft made them widely available within the company.

The development of the MAI models took nearly a year due to technical setbacks, strategy shifts, and the departure of key personnel who disagreed with Suleyman’s management and technical approach. During this period, OpenAI trained and published several batches of advanced models.

Suleyman is focused on preparing Microsoft for AI self-sufficiency in the coming decade, rather than short-term results.

This article was generated with the support of AI and reviewed by an editor. For more information see our T&C.