Microsoft launches faster MAI-Image-2-Efficient for business

Microsoft has rolled out MAI-Image-2-Efficient, a lower-cost and faster version of its in-house image model aimed at teams generating large volumes of visuals for commerce, marketing, UI concepts, and branded assets. The company says the new model is 22% faster than MAI-Image-2, delivers 4x higher efficiency when normalized for latency and GPU usage, and cuts pricing by roughly 41% to $5 per 1 million text input tokens and $19.50 per 1 million image output tokens. Microsoft is positioning it as the production workhorse in the MAI image stack, while the original MAI-Image-2 remains the higher-fidelity option for portraits, deeper photorealism, stylized imagery, and longer in-image text.

The release is targeted at developers and enterprise builders that need speed at scale rather than maximum image polish on every render. Microsoft says MAI-Image-2-Efficient is tuned for real-time and conversational workflows, short-form text rendering such as labels and headlines, and batch pipelines where compute cost matters. In Microsoft Foundry’s model catalog, MAI-Image-2e is listed as a text-to-image model with a 32,000-token context window, PNG output, English support, and configurable width and height, with a minimum size of 768x768 and a maximum pixel budget equivalent to 1024x1024. Regional availability listed by Microsoft includes West Central US, East US, West US, West Europe, Sweden Central, and South India.

This matters because Microsoft is moving fast to turn its MAI model family into a broader platform play inside Foundry. MAI-Image-2 only arrived recently as Microsoft’s flagship image generator, which the company said ranked as the No. 3 image model family on the Arena.ai leaderboard and was already being used by WPP at scale. Microsoft also said MAI-Image-2 had begun rolling out across Copilot, with phased expansion into Bing and PowerPoint. MAI-Image-2-Efficient now gives Microsoft a two-tier image offering: one model tuned for throughput and cost, and another for premium output quality.

Availability starts now in Microsoft Foundry and the MAI Playground, though Microsoft notes the playground remains limited to select markets including the US, with EU countries coming later. The company also says the new model was measured as 40% faster on average than other leading text-to-image systems in its tests, including Gemini-based and GPT-based offerings. Early partner feedback from Shutterstock points to prompt fidelity and production readiness as the main areas where the model is starting to stand out. For Microsoft, the launch is another step in building a first-party stack of speech, voice, and image models that can be sold directly through Azure and wrapped with the governance and deployment controls enterprises already use in Foundry.

Source