Microsoft has released three AI models created in-house, in a move that signals a push toward greater independence from its longtime partner, OpenAI.
Loading audio narration…
MAI-Transcribe-1, MAI-Voice-1, and MAI-Image-2 — which are designed for transcription, voice generation, and image creation — are exclusively available on Microsoft’s Foundry platform, the company announced in a blog post on Thursday.
Microsoft Foundry is a platform within Microsoft Azure, which allows developers and businesses to purchase, build, and customize AI applications using a range of models.
The models are available to enterprise customers, putting them in direct competition with similar tools from OpenAI.
OpenAI offers Whisper for transcription, text-to-speech models for voice, and DALL·E for image generation. All three are also available on Foundry, either as individual products or as part of OpenAI’s broader models.
MAI-Transcribe-1 is “the most accurate transcription model in the world” and MAI-Voice-1 “sets a new standard for natural speech,” said Mustafa Suleyman, CEO of Microsoft AI, in a post on X on Thursday.
Microsoft is OpenAI’s largest investor, and the two companies remain close partners. OpenAI relies on Microsoft’s Azure cloud service to train and run its models, while Microsoft uses OpenAI’s ChatGPT to power Copilot, its AI chatbot.
But Microsoft has increasingly been staking its own claim in the AI race, reducing its reliance on OpenAI by building its own frontier models.
In November, Suleyman formed a superintelligence team at Microsoft focused on training “frontier models of all scales with our own data and compute at the state-of-the-art level” to make the company “self-sufficient in AI,” he told Business Insider at the time.
Previous agreements between the two companies have shaped how independently Microsoft can develop advanced AI systems. A new deal announced in October allowed the companies to “independently pursue AGI (artificial general intelligence) alone or in partnership with third parties.”
