ImageGen2

Garbled lettering is the usual giveaway of an AI-made image, and ImageGen2 is built around fixing that.

ImageGen2 is an AI image generator and photo editor that runs on GPT Image 2, OpenAI's image model, and focuses on near-perfect text rendering, photorealistic output and precise edits.

That matters when the words in the picture are the point: a logo, an infographic, a UI mockup or a product shot with a label that has to be readable.

You can start from a text prompt, or supply up to five reference images for compositing and style transfer.

The same platform handles editing, so you can adjust an existing image instead of regenerating it from scratch.

The model also draws on broad contextual world knowledge, which helps it produce scenes and objects that look plausible.

ImageGen2 is pitched at designers and marketers who need commercial-looking visuals without a photo shoot or a long design cycle.

If your work leans on text inside images, the text-to-image focus on legibility is the main reason to try it, and it covers logos, infographics and realistic photography in one spot.