Microsoft's In-house Model Accelerates Surpass OpenAI, GPU Cost Reduced by Up to 89%

According to Dongcha Beating monitoring, Microsoft AI has released MAI-Image-2.5-Pro and MAI-Voice-2-Flash. The former focuses on high-quality image generation and editing, while the latter targets customer service and voice agents, with a speed twice that of the normal version and a 32% lower price.

Microsoft's proprietary models have also been integrated into Bing, PowerPoint, OneDrive, and Dynamics 365. The GPU cost in PowerPoint is reduced by up to 84% compared to GPT-Image-2, and in Dynamics 365, it is reduced by up to 89%.