Microsoft plans to officially launch several self-developed artificial intelligence models at its annual developer conference Build, which will be held in San Francisco next week. The goal is to further attract the developer community and enhance its own AI system. In response to the current situation where the market share of its code assistant GitHub Copilot is being eroded by Cursor and Claude Code, Microsoft will introduce a cost-effective model specifically designed for coding as an affordable alternative to Claude, aiming to win back price-sensitive developers by reducing operational costs.

Copilot

In addition, Microsoft will also release a series of new iterative models with different parameter specifications, each focusing on popular scenarios such as speech transcription, logical reasoning, speech processing, and image generation, and plans to offer them for purchase through the Azure cloud platform to developers.

This series of strategic moves marks Microsoft's accelerating efforts to reduce its reliance on expensive external technology providers. For a long time, GitHub Copilot and Microsoft 365 Copilot have mainly operated using models from OpenAI and Anthropic. Although Microsoft has free usage rights to OpenAI models until 2032, due to previous partnership agreements, it was unable to develop top-tier models capable of human-level intelligence before the recent contract renegotiation in April, which led to the models developed by the AI chief Mustafa Suleyman's team not ranking at the top of global AI capability lists.

With the expiration of OpenAI's technology usage rights approaching in about six years, Microsoft's move aims to reduce internal operational costs for products like 365 Copilot by focusing on a self-developed model matrix with high cost-effectiveness. This positioning is similar to Google's strategy and also indicates that the global large model market is rapidly shifting from pure capability arms race to deep cost and ecosystem application competition.