Hands-on With 'MAI-Image-1' Image Generator Model; Now Available on Bing

In the past few months, many AI companies launched their improved image generation models, and among them is Microsoft, which actually debuted one. The announcement of the company’s first in-house text-to-image model, MAI-Image-1, came last month.

Per the announcement, the AI image generator model has already grabbed a position in the top 10 leaderboard on LMArena benchmark. So, what’s so special about it? The MAI-Image-1 model focuses on generating more natural and visually rich images.

Now, Microsoft has rolled out the MAI-Image-1 model in the Bing Image Creator interface. The screenshot in the article’s feature image shows that you can select “MAI-Image-1” from the model list alongside DALL-E 3 and GPT-4o. You can try it from here.

The news comes via Mustafa Suleyman on X, who notes that the model “really excels at artistic lighting/photorealistic detail, nature scenes, and food.” Suleyman also notes that MAI-Image-1 is currently available in all countries that can access Bing Image Creator and Copilot Labs, excluding the EU, which is coming soon!

Hands-on With ‘MAI-Image-1’ Model

I tried a few prompts to see how accurate the model can generate images based on text prompt, and I must say I am quite amazed by the quality of the generated images and their details. Take a look below:

Prompt: “Create Halloween pumpkin in a dark jungle setting that looks so scary”

AI-generated Image:

Prompt: “A hyper realistic image of noodles with grated cheese on it, with an omelette on the side, the bowl should be aesthetic design-wise, and in a restaurant setting.”

AI-generated Image:

Prompt: “Cute little puppy playing around its mother, the breed – golden retriever.”

AI-generated image:

Of course, this is still early days. But we want you to try it out now. If you do, we want to hear more about it. You can drop your thoughts about the model below.