Microsoft AI: New Models Challenge OpenAI Dominance

Microsoft has introduced seven new in-house AI models. This announcement came during the company's annual Build Conference. One of these models is MAI-Thinking-1, which is Microsoft's first reasoning model. It's designed for high efficiency and low-token cost. Independent raters preferred MAI-Thinking-1 over Anthropic's Claude Sonnet 4.6 and found it matched Claude Opus 4.6's coding ability in a benchmark. Other new models include MAI-Image-2.5 for image generation, MAI-Transcribe-1.5, MAI-Voice-2, and MAI-Code-1. What's interesting is these new models signal Microsoft's effort to diversify its AI capabilities and reduce its reliance on OpenAI. This comes as their partnership agreement has seen changes, including OpenAI gaining the ability to sell its AI services on competitor platforms like Amazon. The companies also adjusted their revenue-sharing agreement. The bottom line is Microsoft is expanding its own AI offerings, which could impact the broader AI landscape.