ChatGPT will finally let you upload, edit and generate images for free

OpenAI has announced that it will introduce native image generation to ChatGPT, making the feature available to free-tier users. The update, powered by the GPT-4o model, was demonstrated by CEO Sam Altman during a recent live stream. Previously, image generation in ChatGPT relied on OpenAI’s DALL-E text-to-image model, and the new development marks a significant expansion of features for the platform’s free-tier user base.

The announcement comes as part of OpenAI’s ongoing efforts to make AI tools more accessible to a broader audience. If you’re wondering what this brings and what it means for you, read on.

ChatGPT gets native image generation: What's there to know?

According to the company, you will now be able to use the GPT-4o model on ChatGPT to generate, upload, edit and even merge multiple images for your final output. If that’s not all, this enhancement will also mean you can chat with the AI bot to modify your image, across multiple generations, without losing consistency.

Via an announcement on X (formerly Twitter), Altman mentioned that the GPT-4o image generation will be able to carry out the following tasks effectively:

– Render text within images: Allow creation of logos, signs, menus, invites, as well as infographics
– Follow complex prompts: Maintain precision, high fidelity and detailed compositions
– Modify existing images: Build upon a previous output and ensure visual consistency with multiple interactions
– Support multiple artistic styles: From illustrations to photorealism, the new model can carry out the artistic style of your choice

Moreover, users will be able to describe an image to ChatGPT and specify aspect ratio, colour schemes (using hex codes), opacity, and the model will consider it, all within a minute.

ALSO READ: ChatGPT turns 2: How this AI evolved from concept to conversations

This means if you want to create an anime version of yourself, all you’ll have to do is tell ChatGPT of your desired colour palette, art style and upload your image to support this change. This is an incredibly helpful addition for those looking to generate comics, logos and the company even claims GPT-4o can handle up to 20 different objects with a single prompt.

Lastly, the images come with C2PA metadata to help people verify and detect AI-generated images with ease.

ChatGPT image generation: Availability details

ChatGPT’s native image generation will be rolled out to all ChatGPT users. This includes free, Plus, Pro, and Team users starting today. The feature will also be available on Sora; however, Enterprise and Edu users will have to wait a little more.

The feature was demoed back in May 2024 and has only made its way to users after a year. While it remains unclear as to why the brand decided to hold on to it, the new version now offers a much higher image quality and manages to produce more lifelike images (with text). For more announcements by OpenAI or AI-related news, make sure to subscribe to Unboxed by Croma.