Microsoft launches new in-house AI models it says cut costs up to 89% versus OpenAI - VentureBeat

What the report says

VentureBeat reports that Microsoft has introduced two internally developed AI systems, MAI-Image-2.5-Pro and MAI-Voice-2-Flash, as part of a push to lower the expense of running generative AI services. According to the report’s headline and available summary, Microsoft says the models can reduce costs by as much as 89% compared with using OpenAI technology.

The new models are described as intended for Microsoft’s own product ecosystem, including Bing, Excel, Copilot and Dynamics. VentureBeat reports that the effort is aimed at reducing GPU costs, a major expense for companies deploying AI at scale, while also lessening Microsoft’s dependence on OpenAI for some AI capabilities.

The development matters because Microsoft has been one of OpenAI’s most important commercial partners and has built OpenAI-powered features into many of its products. At the same time, major technology companies have been investing in proprietary models to improve control over performance, cost and product integration.

The full VentureBeat article text was not available in the supplied material, so details such as release timing, technical benchmarks, deployment status and how Microsoft calculated the claimed savings could not be independently summarized here. Based on the available evidence, the key point is that Microsoft is expanding its in-house AI model portfolio in areas including image and voice, with cost efficiency and reduced reliance on outside model providers presented as central goals.