Microsoft is replacing OpenAI\u2019s image-generating models with its own technology in popular products like PowerPoint and Bing.<\/p>\n
\u201cIt\u2019s faster, it\u2019s cheaper, it\u2019s higher quality, it drives better retention,\u201d Microsoft\u2019s model chief Mustafa Suleyman said in an interview.<\/p>\n
The world\u2019s largest software maker has been developing artificial intelligence models that can compete with those from OpenAI and Anthropic. In part, this is to drive down the costs of deploying AI across Microsoft\u2019s products. Suleyman said the company\u2019s Microsoft AI (MAI) models are about 85% cheaper to run in PowerPoint, for example.<\/p>\n
For now, the company gets free access to OpenAI models owing to its longstanding partnership. Still, Microsoft must provide the expensive computing infrastructure to run these models, meaning it can realize significant savings by using more efficient ones.<\/p>\n
Customers routinely generate and edit images in Microsoft apps, such as creating infographics in PowerPoint, Suleyman said. Asking the software to create an image is also one of the first AI prompts new users make, meaning it\u2019s important for the experience to be positive, he added.<\/p>\n
Companies such as T-Mobile US Inc. and easyJet PLC are now using Microsoft\u2019s newest voice-generating AI models in their call centers, the company announced. \u00a0<\/p>\n
Bloomberg previously reported that Microsoft was beginning to replace OpenAI and Anthropic text models within widely used office apps. Microsoft said that MAI models in Excel are on-par with OpenAI\u2019s GPT-5.6 for common tasks while being more cost-efficient.<\/p>\n
MAI models are now being used in more than half of all Microsoft products and tested in all of them, Suleyman said. \u201cIt is a very strong showing, and we\u2019re going to continue to push extremely hard,\u201d he said.<\/p>\n
The Reckoning:<\/p>\n
Related stories:<\/p>\n