Microsoft Launches Advanced AI Models Surpassing Competitors
- Microsoft’s MAI-Thinking-1 model outperformed Anthropic’s Claude Sonnet 4.6 in blind evaluations.
- MAI-Image-2.5 models surpassed Google’s Nano Banana Pro on image-editing leaderboards.
- The new models include MAI-Code-1-Flash, MAI Transcribe-1.5, and MAI-Voice-2, supporting multiple languages and tasks.
- MAI-Thinking-1 scored a remarkable 97% on the AIME benchmark for advanced problem-solving skills.
Microsoft unveiled seven new AI models at its Build event, marking a significant step in developing proprietary AI technology beyond its collaboration with OpenAI. These models are designed to enhance reasoning, coding, image editing, transcription, and voice generation capabilities.
The launch of these advanced AI models demonstrates Microsoft’s commitment to leading in frontier AI development by delivering high-performance solutions that meet diverse technological needs while maintaining cost efficiency. (Source)
