Microsoft accelerates image generation with MAI-Image-2-Efficient: 40% faster than the average model

Microsoft has launched MAI-Image-2-Efficienta new, more efficient version of its imaging model that is 40 percent faster than average of text-to-image conversion models in the industry, in addition to reducing costs.

The technology company presented its MAI-Image-2 model in mid-March as an option capable of generate professional photorealistic results from textoffering expressive images and reliable integrated text.

However, just a month later, Microsoft has launched the new version MAI-Image-2-Efficientdesigned for generating images but with a focus on “speed and scalability.” This is because their capabilities are 22 percent faster than the standard version and operates with a cost that is reduced by almost 41 percent.

But above all, the company has stressed that the Efficient version is not only faster than its main product MAI-Image-2, but, according to its tests, Achieves response speeds 40 percent faster than average from other leading text-to-image conversion models, as stated in a statement on its website.

So, while MAI-Image-2-Efficient can offer a response in 13.70 seconds When faced with a standard ‘prompt’, MAI-Image-2 takes around 17.50 seconds, and other models in the sector such as Gemini 3 Pro Image from Google rise to 19.10 seconds o GPT-Image-1.5-High de OpenAI at 41.41 seconds.

Taking all this into account, Microsoft has detailed that MAI-Image-2-Efficient is the “ideal production tool”, since it is intended to be used when a high volume of work, speed and “precise cost control”.

For example, it can be used to generate product photos, marketing creatives, UI mockups, or brand assets. Likewise, the company also highlights its use for short texts such as headlines or labels, in addition to being optimized to run in interactive workflows in real time.

For its part, the model MAI-Image-2 is more focused on a precision tool, Therefore, Microsoft recommends its use when carrying out tasks that require “maximum fidelity.” This is the case of portraits, photorealistic scenes or illustrations. It is also optimal for use with long or complex text embedded in images.

With all this, MAI-Image-2-Efficient is now available on the platforms of Microsoft Foundry y MAI Playground. Likewise, there is also implementing in Copilot and Bing and will soon be available on platforms such as PowerPoint.