Microsoft launches three new foundational AI models

Microsoft announced the release of three foundational AI models on Thursday. Developed by the internal MAI Superintelligence team—an AI research team formed and announced in November 2025—the models are capable of generating text, voice, and images. The launch represents an effort by Microsoft AI to build out its own stack of multimodal models. The models were developed under the leadership of Mustafa Suleyman, the CEO of Microsoft AI.

The new models—MAI-Transcribe-1, MAI-Voice-1, and MAI-Image-2—are being released on Microsoft Foundry, which serves as a platform for releasing AI models. MAI-Image-2 was originally released on March 19 on MAI Playground, a large language model testing software. All three models are now available on the model release platform, and the transcription and voice models have also been added to the testing software. Suleyman indicated that the company will release more models on the platform and directly within Microsoft products soon.

According to a company press release, the models offer specific performance and pricing structures:

• MAI-Transcribe-1: This model transcribes speech across 25 different languages into text. It is 2.5 times faster than Microsoft’s Azure Fast offering, with pricing starting at $0.36 per hour.
• MAI-Voice-1: An audio-generating model that allows users to generate 60 seconds of audio in one second. Pricing starts at $22 per 1 million characters.
• MAI-Image-2: A model with pricing starting at $5 for 1 million tokens for text input and $33 for 1 million tokens for image output.

In an increasingly crowded market, the company wrote in a blog post that MAI hopes a selling point for these models is that they are cheaper than those from Google and OpenAI. Despite releasing its own models, Microsoft remains committed to its partnership with OpenAI, Suleyman reaffirmed in an interview with VentureBeat. Microsoft has invested more than $13 billion into OpenAI and hosts its models across various products. Suleyman told The Verge that a recent renegotiation of that partnership allowed Microsoft to pursue this superintelligence research.

Suleyman emphasized the philosophy behind the new models, stating: “At Microsoft AI, we’re building Humanist AI. We have a distinct view when creating our AI models — putting humans at the center, optimizing for how people actually communicate, training for practical use”

Why it matters

This launch highlights Microsoft’s dual-track strategy: investing heavily in OpenAI while simultaneously building an independent, cost-competitive stack of multimodal models to capture more of the AI market.