OpenAI has been steadily retiring the DALL-E brand, and since April 2026 GPT Image 2 has served as the default image model inside ChatGPT and through the API.
The upgrade brings noticeably stronger photorealism, much better accuracy when following long or complex prompts, and reliable text rendering across multiple languages — historically the weakest point of AI image tools.
It also produces consistent characters and styles across a series of images, which matters for anyone building campaigns, storyboards or product catalogues. Because it lives directly inside ChatGPT, millions of users can now generate professional-grade visuals through plain conversation, with no separate tool or subscription to learn.
Generated images carry content-authenticity tags too, addressing the growing demand for transparency about what is AI-made. For platforms like Aikiros, the takeaway is that users now treat campaign-ready visuals with clean in-image text and visual consistency as the baseline, not a bonus.