Image Gen

A benchmark run in three parts: the prompt for an "Aurelia" logo wordmark on the left, a list of image models in the middle with Nano Banana Pro selected, and the generated AURELIA wordmark on a transparency checkerboard on the right.

Introducing IMG.LY AI Benchmarks

We ran 15 image models through 37 production design prompts and published every score, every prompt and every generated image. Why we built it, why the scores are weighted per job, and the result that surprised us most.

7 min read