Microsoft has launched seven proprietary AI models under the MAI brand, marking a significant step in the company’s move to reduce its dependence on OpenAI and control its own AI technology stack. The models span different capabilities and use cases, with MAI-Code-1-Flash targeting software development tasks and MAI-Thinking-1 designed for complex reasoning.
The most notable aspect of MAI-Thinking-1 is its origin: Microsoft trained the model from scratch using commercially licensed data, with no refinement from OpenAI. This represents a clean separation from the OpenAI pipeline that has underpinned Microsoft’s AI products since the companies deepened their partnership in 2019.
Why Microsoft is building its own models
The practical reasons are clear. Relying exclusively on OpenAI for model capabilities creates dependency on a single external vendor, limits Microsoft’s ability to customize models for specific use cases, and ties Microsoft’s cost structure to OpenAI’s pricing decisions. Building proprietary models gives Microsoft control over all three variables.
There is also a strategic dimension. The relationship between Microsoft and OpenAI has grown increasingly complex as OpenAI has pursued commercial goals that sometimes compete with Microsoft’s own enterprise AI products. Developing MAI models allows Microsoft to serve customers independently of that relationship.
What each model focuses on
MAI-Code-1-Flash is positioned as a fast, cost-efficient model for programming tasks, competing directly with OpenAI’s o-series coding models and with Anthropic’s Claude Sonnet in code-focused applications. The Flash designation signals a focus on inference speed and cost per token rather than maximum capability.
MAI-Thinking-1 is aimed at complex, multi-step reasoning tasks, a category where frontier models have seen the most rapid improvement in recent months. By training this model from scratch on commercially licensed data, Microsoft avoids the licensing constraints that apply to models built on top of OpenAI’s foundation.
What this means for Azure customers
Azure customers have historically accessed OpenAI models through Azure OpenAI Service, which gives Microsoft distribution but not model ownership. MAI models are natively Microsoft’s, which means they can be deployed, priced, and customized without the licensing structure that applies to OpenAI models.
For enterprise customers, this opens the possibility of fine-tuned MAI models for specific industries. For developers and enterprise teams in Latin America and globally building on Azure, MAI models represent a new set of options that will likely become default recommendations for certain workloads.
For sales teams tired of cold leads, slow customer responses, and manual processes, Dapta is the ultimate tool.
Dapta is the leading platform for creating AI sales agents specifically designed to increase inbound lead conversion. Respond to your leads in less than a minute with voice AI and WhatsApp that converts.
If you want your team to sell more while AI handles the complex stuff, you have to try it.
Microsoft’s MAI launch is one of the most consequential AI industry moves of the year: a major platform player asserting model independence and signaling that the OpenAI-centric era of enterprise AI is giving way to a more competitive, multi-model landscape. For enterprises evaluating AI strategy, Microsoft’s move means more choices, more leverage in vendor negotiations, and a more durable Azure platform.
