Industry Odisha Bureau, July 24: In latest news, Microsoft has launched two new in-house artificial intelligence (AI) models in public preview, MAI-Image-2.5-Pro and MAI-Voice-2-Flash. This comes as the company expands its proprietary AI ecosystem and reduces its dependence on OpenAI’s technology.
Microsoft said that MAI-Image-2.5-Pro is its highest fidelity image generation model. It is designed for detailed visual creation, accurate text rendering within images and professional grade editing. The company has already deployed the model across Bing Image Creator, Power Point and OneDrive. It claims that the model reduced GPU serving costs by up to 84% compared with OpenAI’s GPT Image 2.
Meanwhile, MAI-Voice-2-Flash is a high-speed speech model built for enterprise call centres. Priced at $15 per million characters, the model runs twice as fast as its predecessor and powers Dynamics 365 Contact Center. Microsoft said it has reduced operational GPUcosts by up to 89% for customers, including T-Mobile and Easy Jet.
Notably, the launch of Microsoft’s new AI models aligns with CEO Satya Nadella’s “frontier diffusion” strategy. Under this, Microsoft uses smaller, task-specific AI models for routine enterprise workloads while reserving larger frontier models for more complex reasoning tasks.
Further, the company added that these specialised models can run on older Nvidia H100 and A100 GPUs. This will help in preserving advanced computing resources for AI research and training. It is also making the models available to enterprise customers through Azure Foundry. Businesses can fine-tune them using their own data for specific use cases.
Both MAI-Image-2.5-Pro and MAI-Voice-2-Flash are currently available in public preview through the Microsoft Foundry and the MAI Playground.
