Microsoft has just announced three new platformed artificial intelligence models including MAI-Transcribe-1, MAI-Voice-1 and MAI-Image-2, marking an important step forward in its ambition to build its own multi-mode AI ecosystem.
This is a product of Microsoft AI, a Artificial Intelligence research department led by CEO Mustafa Suleyman, established at the end of 2025.
The launch of these models shows that Microsoft is gradually reducing dependence on partners, while directly competing with big names like OpenAI or Google.
In which, MAI-Transcribe-1 is a voice-to-text model, supporting up to 25 languages and is said to be 2.5 times faster than the current Azure Fast service.
MAI-Voice-1 focuses on creating sound, capable of creating 60 seconds of voice in just one second and allows voice customization according to user needs.
Notably, MAI-Image-2 does not only stop at images but also supports video creation, expanding the application capabilities of AI in content creation.
This model has been tested since March 19 on MAI Playground, Microsoft's new model testing platform – before being uploaded to the Microsoft Foundry ecosystem.
Currently, all three models are available on Microsoft Foundry, while voice-related models are also integrated into MAI Playground for testing and development.
According to Mr. Mustafa Suleyman - CEO of Microsoft AI, Microsoft AI's development philosophy is to put people at the center.
The models are designed to optimize the way people communicate in reality, instead of just focusing on technical performance. He also added that many new models will soon be announced and integrated directly into Microsoft products.
Another notable point is the price strategy. Microsoft said that MAI models are priced lower than many competitors. Specifically, MAI-Transcribe-1 costs from 0.36 USD per hour, MAI-Voice-1 from 22 USD per million characters, and MAI-Image-2 costs from 5 USD per million text input tokens and 33 USD per image output.
In the context of the increasingly competitive large language model market, the cost factor is considered an important advantage to attract businesses and developers.
Despite promoting the development of its own models, Microsoft still affirms its continued close cooperation with OpenAI. The company has invested more than 13 billion USD in this partner and integrated many AI technologies into its product ecosystem.
However, recent adjustments in the cooperation agreement have opened up more space for Microsoft to pursue research on "superintelligence". This shows that the company is pursuing a parallel strategy of both cooperation and technological autonomy.
