Microsoft is making a big push into artificial intelligence, going beyond just adding features to existing products. Over the past week, the company launched new upgrades for its Copilot tools and introduced three new AI models built entirely in-house. This move shows Microsoft’s focus on developing the core AI technology that powers its services, giving it more control and flexibility. These new models aim to impact speech, images, and audio, making AI more integrated into everyday business tasks.

Introducing Microsoft’s New Foundation AI Models

Microsoft unveiled three foundational AI models — MAI Transcribe 1, MAI Voice 1, and MAI Image 2 — which are now available through Microsoft Foundry and the new MAI Playground. These tools are designed to help developers test and implement AI features directly. Each model focuses on a different task: MAI Transcribe 1 converts speech into text, MAI Voice 1 generates spoken audio from text, and MAI Image 2 creates images from prompts. These capabilities are especially useful for meetings, customer support, media creation, and office productivity tasks.

This launch marks a shift for Microsoft. The company is no longer just a provider of AI access through Azure and Copilot. Instead, it’s building more of the underlying AI technology itself. This gives Microsoft better control over costs, speeds up product development, and improves the quality of its AI features. It’s also a strategic move to show investors that AI investments can turn into real revenue streams, especially as the company faces pressure from recent stock performance and economic challenges.

What the New AI Models Do in Practice

MAI Transcribe 1 is the flagship model of this release. It offers high-accuracy speech-to-text conversion across 25 languages. Microsoft says it’s already being tested inside Copilot Voice mode and Microsoft Teams. Importantly, it uses roughly half the GPU resources compared to other top models, which matters because transcription is a daily task for many businesses and happens at high volume. This efficiency can significantly reduce costs and improve performance for large-scale deployments.

MAI Voice 1 is Microsoft’s latest text-to-speech model. It can generate 60 seconds of audio in less than a second on a single GPU. Companies can also create custom voices from short audio samples using Foundry, making it easier to build branded voice assistants, support systems, and media products. This customization opens up new possibilities for personalized customer interactions and media creation.

Meanwhile, MAI Image 2 is an upgraded image generation model. Microsoft reports that it ranks among the top three in the Arena.ai leaderboard for image models. It now runs at least twice as fast as the previous version on Foundry and Copilot. Microsoft is rolling out this model across Bing and PowerPoint, enhancing visual content creation and presentation design with AI-generated images. The improvements in speed and quality mean more dynamic and creative outputs for users.

Overall, these new AI models reflect Microsoft’s broader strategy to develop and control the core AI building blocks. By doing so, it can better tailor features, reduce costs, and accelerate deployment across its product ecosystem. This move also helps Microsoft position itself as a leader in AI innovation, competing more directly with other tech giants investing heavily in foundational AI technology.

Inspired by

• https://justainews.com/companies/microsoft/microsoft-new-ai-models-push-microsoft-ai-beyond-copilot/