Microsoft unveils new AI models to lessen reliance on OpenAI and lower costs for developers Microsoft has introduced its first proprietary AI models, MAI-Code-1-Flash for generating code and MAI-Thinking-1 for reasoning, aiming to compete with leading AI providers and reduce costs for developers. These new models are designed for efficiency and lower token costs, allowing Microsoft to leverage its Azure cloud infrastructure. The company seeks to move beyond just consuming frontier models to actively participating in the AI ecosystem, with MAI-Code-1-Flash integrated into GitHub Copilot and Visual Studio Code.

• Microsoft announced MAI-Code-1-Flash, its first model that generates source code from text descriptions.
• MAI-Thinking-1, a new reasoning model, is also being introduced with a focus on efficiency and low token costs.
• These proprietary models aim to reduce developer costs associated with using third-party AI services like OpenAI’s.
• Microsoft is making these models available through its Azure cloud infrastructure.
• MAI-Thinking-1 is currently in private preview via Microsoft Foundry.
• The coding model is integrated into GitHub Copilot and Visual Studio Code.
• Microsoft also revealed updated cloud models for speech and image generation, and small AI models for Windows PCs.