Microsoft launches its own AI models and starts reducing dependence on OpenAI

Microsoft used Build 2026 to unveil a new family of in-house artificial intelligence models, including MAI-Thinking-1, its first advanced reasoning model. The move matters because it shows Microsoft strengthening its own AI stack, even while keeping its strategic relationship with OpenAI.

For years, Microsoft has been seen as OpenAI’s biggest partner. Now, the message is shifting: Microsoft wants its own AI “brains” powering Windows, Copilot, GitHub, and enterprise tools.

Quick answer: what is happening?

Microsoft launched its own models for reasoning, coding, image generation, voice, and transcription. The biggest highlight is MAI-Thinking-1, an internally trained reasoning model that reinforces Microsoft’s plan to rely less exclusively on OpenAI for its AI roadmap.

What is MAI-Thinking-1?

MAI-Thinking-1 is Microsoft AI’s first advanced reasoning model. It is designed to handle tasks that require planning, logic, and multi-step problem solving.

This type of model matters because AI is moving beyond simple question-and-answer use cases. The next battle is about systems that can reason better, code better, analyze context, and execute complex workflows with more consistency.

Microsoft wants its own AI foundation

The most interesting part is not just the model itself. It is the strategy.

Microsoft already has Copilot across Windows, Office, Teams, GitHub, Azure, and Edge. To power all of that, depending only on external models could become a technical and business risk.

With its own models, Microsoft gains more control over cost, performance, integration, and product speed.

This is not a breakup with OpenAI

This does not mean Microsoft and OpenAI are splitting. Their partnership remains important, especially around Azure infrastructure and products already using OpenAI technology.

But the balance is changing. Microsoft appears to be building a more hybrid strategy: use partners when it makes sense, while also owning more of the technology itself.

Code, voice, image, and transcription

Alongside MAI-Thinking-1, Microsoft also introduced models for image generation and editing, voice, transcription, and coding. One highlight is MAI-Code-1-Flash, a coding model integrated with GitHub Copilot and Visual Studio Code.

That shows Microsoft is not just building another chatbot. It is building a full AI layer for productivity, development, and creation.

Why this matters to you

For regular users, this could mean a faster, smarter, and more deeply integrated Copilot inside Windows and Microsoft apps.

For developers, it could mean more specialized coding tools built directly for the Microsoft ecosystem.

For the market, the message is clear: the AI race is entering a new phase where big tech companies do not want to depend on a single model provider.

Frequently asked questions

Is Microsoft leaving OpenAI?

No. Microsoft still has a strategic relationship with OpenAI, but it is strengthening its own model lineup to reduce dependence.

What is MAI-Thinking-1?

It is Microsoft AI’s first advanced reasoning model, designed for more complex, multi-step tasks.

Will these models come to Windows?

The trend is that parts of this technology will be integrated into Copilot, Windows, GitHub, and Microsoft’s enterprise tools.

Does this change the AI race?

Yes. It shows the battle is no longer just between OpenAI, Google, and Anthropic. Microsoft wants to own core models in the next phase of AI.

At DigitalRadar, this is a major signal: Microsoft does not want to just distribute AI. It wants to build the intelligence behind it.