Microsoft introduced new MAI models, improving performance

Microsoft has announced three new AI models: MAI-Transcribe-1, MAI-Voice-1, and MAI-Image-2. These models are available from Microsoft Foundry and will be implemented in a variety of consumer products. MAI-Transcribe-1 excels in text transcription, purportedly outperforming models from rivals such as Google and OpenAI. Meanwhile, MAI-Voice-1 is known for its ability to produce natural speech with emotional depth. The models concentrate on internal improvements aimed at increasing performance.

Availability Details

Microsoft Foundry and MAI Playground have begun offering MAI models, including MAI-Transcribe-1, available to all developers. Users may run these models using the MAI Playground, which is presently only accessible in the United States.
MAI-Transcribe-1 costs $0.36 per hour; MAI-Voice-1 is $22 per million characters; and MAI-picture-2 charges $5 per million tokens for text input and $33 per million tokens for image output. Additionally, Microsoft Foundry and Copilot are demonstrating these models for both business and consumer applications.

Microsoft’s New AI Models

Microsoft announces three cutting-edge models: MAI-Transcribe-1, MAI-Voice-1, and MAI-Image-2, all of which provide remarkable quality at low prices. MAI-Transcribe-1 provides superior speech-to-text transcription capabilities in 25 languages, with a transcription speed that is 2.5 times faster than previous Microsoft Azure offerings. This model is particularly designed for accuracy in dynamic real-world scenarios and boasts the highest price-performance ratio among major cloud providers.

Also Read: OnePlus 15R price in India increased by Rs 2,500

MAI-Voice-1 specialises in natural voice generation, enabling you to create custom voices with only a few seconds of audio input. It is very efficient, creating 60 seconds of audio in one second while optimising GPU utilisation to enable the creation of varied voice experiences.

Also Read: Google introduced Gemma 4, an open-source AI model, bringing agentic features and advanced reasoning to the open-source model

MAI-Image-2 significantly improves picture-generating speed and performance, with generation times that are at least twice those of comparable models. It is designed specifically for photographers and designers, ensuring the generation of photographs with natural lighting and precise textures while being competitively priced.

Furthermore, Microsoft Foundry and Copilot highlight these models for a wide range of commercial and consumer applications, emphasising their adaptability and advanced technology.