Microsoft Debuts AI Models to Cut Developer Costs

Microsoft has introduced two proprietary AI models at its Build 2026 conference in San Francisco, moving to reduce its dependence on third-party providers like OpenAI.

This marks a notable shift for Microsoft, which has largely served as an AI infrastructure provider and investor. Building its own models lets Microsoft avoid paying third parties and pass lower costs to developers.

The New Models

At Build 2026 on June 2, Microsoft announced MAI-Code-1-Flash, its first AI coding model. The model takes written text descriptions and generates source code for applications and websites.

Microsoft also introduced MAI-Thinking-1, a reasoning model. Kyle Daigle, Microsoft's developer marketing chief and GitHub operating chief, described it as "built for high efficiency and performance, but importantly, at a low-token cost." Tokens are the units of data a model processes, and their volume directly determines what developers pay.

Performance and Availability

Microsoft says MAI-Thinking-1 outperformed OpenAI's GPT-5 after being refined for consulting firm McKinsey, with 10 times better cost efficiency, according to Mustafa Suleiman, CEO of Microsoft AI.

MAI-Code-1-Flash is already available in GitHub Copilot and the Visual Studio Code editor. MAI-Thinking-1 is in private preview through Microsoft Foundry, and customers can express interest in testing it before it becomes broadly available. Microsoft also said customers will be able to improve the reasoning model's accuracy by incorporating their own data.

The Business Context

Microsoft has invested $13 billion in OpenAI and $5 billion in Anthropic, and currently offers both companies' models through its Azure cloud platform. Its primary role in the AI sector to date has been as a cloud infrastructure provider and equity investor.

The new models represent an effort to compete directly with the providers it funds. Anthropic filed confidentially for an IPO on June 1, and OpenAI is also pursuing a public offering, potentially in 2026.

CEO Satya Nadella framed the shift onstage at Build: "What you just saw is a pretty significant shift. We believe the time has come for every company to just move from consuming a frontier model to fully participating at the frontier in the frontier ecosystem."

Additional Announcements

Alongside the two main models, Microsoft unveiled updated cloud-based tools for speech recognition, synthetic voice generation, and image generation. The company also announced small Aion models designed to run directly on Windows PCs.

Google announced its Gemini 3.5 Flash model in May 2026, which also handles coding and runs within the company's own data centers, indicating similar cost-reduction moves across the industry.

About Microsoft

Microsoft is a global technology company headquartered in Redmond, Washington. It develops software, cloud services, and devices, and operates Azure, one of the world's largest cloud computing platforms. The company has made major investments in AI through stakes in OpenAI and Anthropic, and sells access to both companies' models through its Azure AI services.