Microsoft AI (MAI) has introduced two new models as part of its goal to build “AI for everyone” — models that are responsible, reliable, and useful for daily life.

The first, MAI-Voice-1, is a speech generation model that produces high-quality, expressive audio. Already powering Copilot Daily and Podcasts, it can also be tested in Copilot Labs through demos like storytelling and meditation. The model generates one minute of audio in less than a second on a single GPU, supporting both single- and multi-speaker use cases. Microsoft called voice the “interface of the future” for AI companions.

The second, MAI-1-preview, is Microsoft’s first foundation model, now open for testing on LMArena. Built with around 15,000 NVIDIA H100 GPUs, it is trained for instruction-following and conversations. It will soon power text-based features in Copilot, with public feedback guiding improvements. Developers can also apply for API access.

Microsoft said these launches are the first steps in a bigger roadmap, aiming to combine specialised AI models with contributions from partners and the open-source community. The company believes this approach will unlock significant value and bring advanced AI into the hands of people worldwide.

Tired of missing hot stocks? Unicorn Signals provides powerful tools like stock scans and more help you make informed trading decisions. Download now and take control of your portfolio!