AI Microsoft was announced His first home AI models on Thursday: Mai-Voice-1 AI and Mai-1-Review. The company claims that its recent Mai-Voice-1 speech model can generate a minute of audio in a smaller second on one GPU, while Mai-1-Review “offers a look at future offers in Copilot”.
You can try MA1-Voice-1 for yourself at Copilot LabsWhere you can introduce what you want to say AI model, as well as change his voice and style of speaking. In addition to this model, the Microsoft was introduced by Mai-1-Review, which, he claims, trained about 15,000 GPU NVIDIA H100. It is built for users in need of the AI model, which is able to follow the instructions and “ensuring helpful answers to daily queries.”
Boss of Microsft AI Decoder Last year, internal AI models of the company do not focus on cases of operate of enterprises. “My logic is that we have to create something that works extremely well for the consumer and really optimize for our case of use,” said Suleyman. “So we have huge amounts of very predictive and very useful data on the advertising side, on consumer telemetry and so on. I focus on building models that really work for a consumer companion.”
Microsoft AI plans to introduce Mai-1-P-PreView for some cases of using the text in its assistants Copilot AI, which is currently based on gigantic models of the OpenAI language. He also began to publicly test his model Mai-1 P-Review on the AI Lmarena comparative platform.
“We have great ambitions to Where, we go further,” writes Microsoft AI in the blog post. “Not only will we continue to continue here, but we think that organizing a number of specialized models that support various user intentions and cases of use will unlock enormous value.”