Microsoft AI (May) started public tests from the first foundation model he formed internally.
The model, nicknamed Mai-1-Preview, is tested on LMARENA, a platform for the assessment of community models, said the company on Thursday, August 28) blog.
“This represents the first May foundation model formed from start to finish and offers an overview of future offers inside Copilot,” said the company in the position. “We actively turn the steering wheel to provide improved models.”
According to the Post, Mai-1-Preview is designed to be used by consumers and specializes in the following instructions and the answer to daily questions.
It will be deployed for certain cases of text use in Copilot in the coming weeks, according to the post.
“We will continue to use the best models of our team, our partners and the latest innovations in the open source community to feed our products,” said May in the position. “This approach gives us flexibility to provide the best results in millions of unique interactions every day.”
CNBC reported Thursday that Microsoft feeds the characteristics of the artificial intelligence of its bing search engine, its Windows 11 operating system and other products mainly with models of AI of OPENAI– A company in which Microsoft has invested more than $ 13 billion – and that the development of an internal model could point out that it works to reduce this dependence.
Microsoft added Openai to a list of competitors in its annual report last year, while Openai added cloud suppliers beyond Microsoft, including Coreweave, Google and Oracle, according to the report.
May also announced in his Thursday blog article that he published a natural speech generation model called Mai-Voice-1, which makes him available in Copilot Daily and Podcasts and as an experience of Copilot Labs.
“La Voix is the interface of the future for the AI and Mai-Voice-1 companions provides an expressive sound with high fidelity in unique and multi-storage scenarios,” said May in the post.
This announcement came the same day that Openai published what he calls his “most advanced discourse model to date”. The company has also created its general -time API application programming interface generally available, claiming that the application programming interface now has features that help developers build voice agents.
