Microsoft Begins Public Testing of In-House AI Model MAI-1-preview

Aug 29, 2025
Microsoft has started public testing of its first in-house AI model, MAI-1-preview, on the LMArena platform, marking a step towards reducing reliance on OpenAI models.

Microsoft has initiated public testing of its first internally developed AI model, MAI-1-preview, on the LMArena platform. This marks a significant step for Microsoft AI (MAI) as it seeks to reduce its reliance on models from OpenAI, a company in which Microsoft has invested over $13 billion. The MAI-1-preview model is designed to handle text-based tasks and will be integrated into Microsoft's Copilot for specific use cases in the coming weeks.

The model, which has been trained using approximately 15,000 Nvidia H100 GPUs, is part of Microsoft's broader strategy to develop its own AI capabilities. This move could potentially escalate competition with OpenAI, as Microsoft aims to enhance its AI offerings across products like Bing and Windows 11.

In addition to MAI-1-preview, Microsoft has also introduced MAI-Voice-1, a natural speech generation model available in Copilot Daily and Podcasts. This model is designed to deliver high-fidelity, expressive audio and is part of Microsoft's vision for AI companions that can interact naturally with users.

We hope you enjoyed this article

Consider subscribing to one of our newsletters like Daily AI Brief.

Also, consider following us on social media:

AI Brief AI Brief (X)
Free newsletter

Daily AI Brief

Daily report covering major AI developments and industry news, with both top stories and complete market updates

AI Token Webinar

Whitepaper

Tensordyne Napier: What If One Rack Could Do the Work of Nine?

Tensordyne

This Tensordyne whitepaper presents Napier, an inference-focused AI processor and rack-scale system based on the company’s TDN Math logarithmic number system. It examines infrastructure requirements for large mixture-of-experts and agentic models, compares major inference architecture approaches, and details the TDN AIP processor, TDN72 pod, TDN Link fabric, and Napier Ultra configuration. The paper reports simulation-based performance, cost, and accuracy-validation results, including Tensordyne’s projected comparison of one Napier rack with a nine-rack Nvidia Rubin plus Groq deployment; the chip is reported as taped out and in fabrication.

Read more

You may also like

Free, six days a week

Daily AI Brief: the AI news that matters, in your inbox.