Microsoft expanded its in-house artificial intelligence ecosystem on July 23, 2026, by launching public previews for MAI-Image-2.5-Pro and MAI-Voice-2-Flash. The models reduce reliance on external suppliers like OpenAI and have already been deployed across PowerPoint, Bing, and OneDrive.
Microsoft is quietly executing a foundational shift across its software empire, steering workloads away from external partners like OpenAI and Anthropic and toward its own family of in-house models. Chief Executive Officer Satya Nadella outlined the strategy in a post on X, explaining that software now carries a real marginal cost for the first time. That economic reality makes model selection a core business decision rather than a technical default. Frontier models remain in place for complex problems, but routine tasks are increasingly handled by cheaper, purpose-built alternatives. Satya Nadella says Microsoft is now routing tasks inside GitHub Copilot, Excel and Outlook to its own MAI models whenever they match or beat frontier alternatives from OpenAI and Anthropic.
Deploying MAI Models Across Office, Bing, and Azure
The push is no longer confined to research laboratories or limited testing environments. Microsoft has embedded its MAI models directly into major consumer and enterprise applications. Bing Image Creator now runs entirely on MAI-Image-2.5 end-to-end by default, replacing OpenAI’s image-generating models. Meanwhile, PowerPoint utilizes the model for image-to-image work, and OneDrive relies on it for key editing tasks. Microsoft Corp. is replacing OpenAI’s image-generating models with its own technology in popular products like PowerPoint and Bing.
The efficiency gains from these internal deployments are stark. According to production metrics released by the company, switching to the MAI image stack in PowerPoint reduces GPU costs by up to 84% compared with OpenAI’s GPT-Image-2. In OneDrive, the transition delivered a 26% rise in save success rates, lowered P95 latency by roughly 25%, and increased medium-load production efficiency by 2.5 times.
Public Previews for MAI-Image-2.5-Pro and MAI-Voice-2-Flash
Expanding on its existing architecture, Microsoft opened public previews for two specialized models through Microsoft Foundry and the MAI Playground. MAI-Image-2.5-Pro is engineered for hero artwork, detailed editing, and precise text rendering inside generated images—an area that has historically challenged generative media models. MAI-Image-2.5-Pro is Microsoft’s highest-fidelity image model to date, built for hero artwork, detailed editing and precise text inside generated images, with natural-language commands supporting rapid revisions. WPP Global Chief Creative Officer Rob Reilly called the text rendering a breakthrough and noted that natural-language editing accelerates creative iteration.
On the opposite end of the cost curve, MAI-Voice-2-Flash targets high-volume enterprise voice applications. The speech model runs twice as fast as MAI-Voice-2 and is 32% cheaper while preserving natural prosody and acoustic quality. MAI-Voice-2-Flash is 2x faster than MAI-Voice-2 and 32% cheaper, at $15 per 1M characters. It powers Dynamics 365 Contact Center for enterprise clients such as T-Mobile and EasyJet, achieving GPU cost reductions of up to 89% in contact-center deployments.
“Compared to OpenAI models, our own models run faster, cost less, and deliver higher generation quality, which improves user retention rates.”
Mustafa Suleyman, CEO of Microsoft AI
Reinforcement Learning Gyms and the Path Forward
Behind the hardware savings is a deliberate architectural philosophy. Rather than training models in isolation, Microsoft trains them inside active product environments—what Mustafa Suleyman calls training gyms—where systems evolve through ongoing evaluations driven by customer interactions. By keeping memory, context, tools, and skills outside the models themselves, Microsoft ensures it can upgrade or swap underlying models without disrupting the user experience. As next-generation computing clusters go into operation and the MAI series continues to expand, Microsoft is also commercializing the approach.
