Microsoft Launches Three Models to Reduce Dependence on OpenAI

Microsoft has launched three new MAI models—MAI-Transcribe-1, MAI-Voice-1 and MAI-Image-2—on Microsoft Foundry, adding to its portfolio of AI models for developers and enterprises.

All three models are available through Microsoft Foundry and the MAI Playground in the US.

“At Microsoft AI, we’re building Humanist AI. We have a distinct view when creating our AI models — putting humans at the centre, optimising for how people actually communicate, training for practical use,” said Mustafa Suleyman, CEO of Microsoft AI.

The company said MAI-Transcribe-1 delivers speech-to-text transcription across 25 languages and achieves a 3.9% average word error rate on the FLEURS benchmark. It also processes batch transcription at 2.5 times the speed of Microsoft Azure’s existing fast offering.

MAI-Voice-1 is a voice generation model that allows users to create custom voices from short audio samples. It can generate up to 60 seconds of audio in a single second, supporting long-form speech while preserving speaker identity.

MAI-Image-2 targets image generation and is being rolled out across Copilot, Bing and PowerPoint. Microsoft said the model delivers at least twice the image-generation speed of earlier versions, based on production data. It serves use cases such as marketing, design and visual content creation.

The models are priced as follows: MAI-Transcribe-1 starts at $0.36 per hour, MAI-Voice-1 at $22 per one million characters, and MAI-Image-2 at $5 per one million text tokens and $33 per one million image tokens.

Microsoft added that the models include built-in safeguards, governance controls and testing processes to support enterprise deployment.

The launch of the new models comes as the company looks to reduce its reliance on OpenAI for foundation models. The relationship has been strained in recent months, particularly after OpenAI signed multiple agreements with Amazon, including a deal with Amazon Web Services, which will serve as the exclusive third-party cloud provider for OpenAI Frontier, its enterprise platform.