Microsoft’s MAI-Image-2-Efficient Cuts Costs 41% as Company Reduces OpenAI Dependence

Quick Facts

MAI-Image-2-Efficient costs $5 per million text tokens and $19.50 per million image tokens, a 41% reduction from the flagship model
The new model runs 22% faster than MAI-Image-2 and delivers 4x greater throughput efficiency per GPU on NVIDIA H100 hardware
Microsoft’s AI chief Mustafa Suleyman states the company’s mission is building superintelligence with independent foundational models

Microsoft launched MAI-Image-2-Efficient with significant cost reductions aimed at high-volume enterprise workflows. The model costs $5 per million text input tokens and $19.50 per million image output tokens.

The efficient variant delivers 22% faster performance than Microsoft’s flagship MAI-Image-2 model. Internal testing shows 4x greater throughput efficiency per GPU when measured on NVIDIA H100 hardware at 1024×1024 resolution.

Microsoft designed the model for “assembly line” tasks including user interface mockups, product photography and marketing assets. E-commerce platforms, media companies and marketing teams generating thousands of images daily represent target customers.

The launch accelerates Microsoft’s strategic shift away from OpenAI dependence. MAI models now serve as the default option when users ask Copilot to generate images, replacing OpenAI’s DALL-E model. This change allows Microsoft to retain all costs internally rather than paying licensing fees to OpenAI.

“My personal mission at Microsoft is to build superintelligence,” said Mustafa Suleyman, Microsoft’s AI chief, in a Financial Times interview. “We must develop our own cutting-edge foundational models, equipped with gigawatt-scale computing power, and some of the world’s best AI training teams.”

Microsoft’s flagship MAI-Image-2 reached third position on the Arena.ai leaderboard for image generation with 1326 Elo points. The model outperformed Google’s Gemini Pro 1.5 and Anthropic’s Claude 3 Opus in direct comparisons.

The rapid development timeline stands out. MAI-Image-2 debuted on MAI Playground March 19, gained broader availability through Microsoft Foundry April 2, then shipped an optimized production variant less than a month later.

Microsoft faces OpenAI dependency concerns after analysts flagged that OpenAI represents 45% of the company’s cloud backlog. Following earnings discussions about this exposure, Microsoft shares lost $357 billion in market value in a single day.

The October 2025 partnership restructuring converted Microsoft’s profit-sharing rights into a 27% ownership stake in OpenAI Group PBC, valued at $135 billion. The restructuring gave both companies freedom to operate independently while extending Microsoft’s IP rights to OpenAI models until 2032.

Microsoft plans infrastructure scaling 6 to 10 times beyond current capacity in 2026, backed by over $120 billion in committed capital expenditure. The company has invested over $13 billion in OpenAI since 2019.

This article was written by an AI agent. Spotted an error? Send a correction and we will fix it.