Microsoft’s AI division has unveiled three new foundational AI models designed to enhance text, audio, and image generation capabilities. This release, announced on Thursday, reflects Microsoft’s ongoing commitment to expanding its multimodal AI solutions and competing with leading AI developers, despite its ongoing collaboration with OpenAI.
The new models are as follows:
-
MAI-Transcribe-1: This model efficiently transcribes speech into text in 25 languages at a speed 2.5 times faster than Microsoft’s Azure Fast service.
-
MAI-Voice-1: An innovative audio generation model that creates 60 seconds of audio in just one second and enables users to develop custom vocal profiles.
- MAI-Image-2: This model primarily focuses on video generation and was previously introduced in the MAI Playground, a platform for testing large language models.
These foundational models are now accessible on Microsoft Foundry, and the transcription and voice capabilities can be explored in the MAI Playground.
The development of these models was spearheaded by Microsoft’s MAI Superintelligence team, led by Mustafa Suleyman, who has articulated a vision for "Humanist AI." The aim is to centre AI design around human interaction and communication, ensuring the models are practical and user-focused. Suleyman has indicated that more models will be forthcoming on both Foundry and Microsoft’s product suite soon.
In a competitive landscape dominated by entities like Google and OpenAI, Microsoft is positioning its models as cost-effective alternatives. Pricing for the new models includes MAI-Transcribe-1 starting at $0.36 per hour, MAI-Voice-1 at $22 per million characters, and MAI-Image-2 priced at $5 per million tokens for text input and $33 for image output.
Despite the launch of these proprietary models, Suleyman reaffirmed Microsoft’s dedication to its relationship with OpenAI, detailing that recent renegotiations of this partnership have enabled more aggressive exploration of superintelligent research paths. Microsoft’s substantial investment of over $13 billion into the AI sector demonstrates its commitment to integrating these advanced models across its product ecosystem, while concurrently developing internal capabilities and sourcing from external partners as needed.
As the AI landscape evolves, Microsoft continues to forge a path of innovation, positioning itself strategically among top tech competitors while enriching its suite with user-friendly, efficient AI solutions.
Fanpage: TechArena.au
Watch more about AI – Artificial Intelligence
