OpenAI has launched in-painting features in GPT-4o, enabling users to create visuals within the AI model itself. This tool, launched on Tuesday, can be employed to develop diverse images, including infographics, memes, menus, comic strips, signboards, and others.
Users can fine-tune and edit produced images through follow-up prompts, giving more control.The feature is already rolled out to Free, Plus, Pro, and Team plan subscribers, while Enterprise and Edu plans will follow soon. Also, API integration will be rolled out during the following weeks.
No Longer DALL-E Limited
In contrast to earlier models, GPT-4o can create images on its own from its internal knowledge base and not through diffusion models like OpenAI’s DALL-E.Nevertheless, users can still access DALL-E for generating images when needed. OpenAI elaborated, “Creating and tailoring images is as easy as conversing with GPT‑4o – just tell us what you want, including any details like aspect ratio, precise colors in terms of hex codes, or a transparent background.”
Social Media Responses
GPT-4o’s new capabilities were tested by users straight away, with responses posted online. Shopify CEO Tobias Lütke was stunned when the model accurately described the anatomy of a foreign creature on his son’s t-shirt, answering with, “How is this even real?”
How is this even real?
OpenAI cooked pic.twitter.com/RfRJhv8uFb
— tobi lutke (@tobi) March 26, 2025
Indian users also experimented with the potential of the model by creating images, converting existing pictures into various styles, and even creating user interfaces from text inputs.
Take a look:
can I just say @OpenAI did it better? this is with the gpt4o image generation which is launched yesterday ig. https://t.co/BDsLqlTX48 pic.twitter.com/PERpaYTrbE
— pranav⚡ (@i_pranavvvvv) March 26, 2025
OpenAI knows how to ship products optimized for user delight!
PS – This is image generation with ChatGPT in Studio Ghibli art. pic.twitter.com/BH7CGlxDip
— Aniket Kamthe (@aniketkamthe777) March 26, 2025
🎨✨ Just tried the new OpenAI 4o image generation—and wow.
Fluid forms, dreamlike detail, and conceptual clarity in every curve
This is a huge leap forward for GPT‑4o image generation!
Swipe to see my organic architecture tower concept : sketch + real pic.twitter.com/xsILVRXZB7
— Matt Menendez (@mezdez) March 26, 2025
OpenAI vs. Google in AI Image Generation
OpenAI’s announcement follows Google’s introduction of native image generation in its Gemini 2.0 Flash AI model. Initially tested by select users in December, the feature is now available in all supported regions via Google AI Studio.
Google stated, “Developers can now test this new capability using an experimental version of Gemini 2.0 Flash (gemini-2.0-flash-exp) in Google AI Studio and via the Gemini API.”
