Microsoft Unveils First In-House Image Generator MAI-Image-1, Marking Shift Away From OpenAI Dependence

Microsoft Corp. on Monday announced its first image generation model, MAI-Image-1, designed and developed in-house.

After relying on OpenAI's technology for years, the Windows maker is now pushing its own large language models (LLMs) in the market. That allows Microsoft to deliver AI capabilities through a fully self-owned technology stack.

CEO Satya Nadella reportedly said at town hall last month that he's "looking forward to us building model capability, so that we can build model-forward products." The company announced its first homegrown AI models on Aug. 29: MAI-Voice-1 AI and MAI-1-preview.

On Stocktwits, the retail sentiment for Microsoft's stock was 'bearish' as of late Monday, compared to 'neutral' on Saturday. MSFT gained 0.6% on Monday and is up 20.8% year-to-date.

About MAI-Image-1, Microsoft said it has trained the model to deliver genuine value for creators, avoiding repetitive or generically-stylized outputs, according to a blog post published on Monday.

"For example, we prioritized rigorous data selection and nuanced evaluation focused on tasks that closely mirror real-world creative use cases – taking into account feedback from professionals in the creative industries,' the blog post read.

MAI-Image-1 excels at generating photorealistic imagery, with better lighting effects and other enhancement features. It is currently available on LMArena and will soon be available through Copilot and Bing Image Creator, according to Microsoft.

The move is significant as it is another step in Microsoft’s strategy to reduce its reliance on OpenAI, an investee company, cloud customer, and AI technology partner for Microsoft. Development of in-house LLMs started after Microsoft began adding third-party AI models, from makers such as Mistral, Anthropic, and xAI, to its Azure cloud platform.

For updates and corrections, email newsroom[at]stocktwits[dot]com.

Share

·

Add us on