Microsoft's New In-House AI Models Slash GPU Costs by Nearly 90 Percent
Microsoft is making a massive leap toward self-sufficiency with new internal models designed to deliver high-performance AI at a fraction of the traditional price.
By Blip Tech
1 min read
Microsoft has unveiled two major in-house AI models, MAI-Image-2.5-Pro and MAI-Voice-2-Flash, designed to drastically reduce operational expenses and reliance on external partners like OpenAI. These models, now in public preview, offer up to an 89% reduction in GPU costs while powering core applications such as Bing, Excel, Copilot, and Dynamics. MAI-Image-2.5-Pro stands as the company's most advanced image generator, while MAI-Voice-2-Flash provides high-speed speech synthesis. This strategic shift highlights Microsoft's commitment to optimizing its own infrastructure and tailoring AI performance specifically for its vast ecosystem of productivity tools.
