- calendar_today August 10, 2025
The ChatGPT interface now includes image generation thanks to the new OpenAI upgrade called “Images in ChatGPT.” The feature enables users to generate images during conversations through the capabilities of the new GPT-4o model, which advances AI content creation.
All ChatGPT subscription levels, such as Plus, Pro, Team, and the free version, now include access to the new functionality. The feature’s wide availability seeks to make advanced image generation accessible to everyone. OpenAI spokesperson Taya Christianson said that the current limit of approximately three images per day for free tier users may be subject to change depending on demand. A custom GPT will continue to offer a pure DALL-E experience for those users who desire it.
OpenAI research lead Gabriel Goh labeled GPT-4o as an “omnimodal” model, which demonstrates transformative capabilities by processing text, images, audio, and video. The model’s ability to maintain “binding” between objects and attributes has been significantly improved. The model now solves a frequent difficulty in AI image creation, where earlier versions found it hard to preserve correct object-attribute relationships. The GPT-4o system demonstrates a remarkable ability to accurately handle 15 to 20 objects without mixing their colors or shapes.
The system now offers superior performance in text rendering, which stands out as one of its main improvements. AI-generated images are used to display text that was often muddled and meaningless. Goh explained that they performed a lengthy iterative development process that spanned many months until they reached perfection. The team has managed to create consistently usable text within images, even though achieving perfect small text rendering is still difficult.
Instead of using standard diffusion models for image generation, the system employs an autoregressive architecture. The sequential image generation approach from left to right and top to bottom, similar to text creation, may enhance the system’s text rendering and binding performance.
OpenAI demonstrated a variety of applications for the system at a briefing, which include producing scientific diagrams like Newton’s prism experiment with precise labels, along with creating multi-panel comics featuring consistent characters and dialogues, and designing informational posters with precise text. The system showcased practical applications by demonstrating transparent background image generation for stickers, as well as restaurant menus and logos.
As the multimodal product lead for ChatGPT, Jackie Shannon highlighted how the system effectively utilizes world knowledge. She explained that when she creates images with her own skill set boundaries, but she uses her extensive world knowledge as well. The model incorporates world knowledge into its operations, which allows users to request images of Newton’s prism experiment without needing to provide any additional information about the experiment.
OpenAI argues that the improved image quality and added capabilities make the slightly longer generation time worthwhile. Shannon admitted that latency improvement is necessary but stated that image quality and world knowledge capabilities make the wait time worthwhile.
OpenAI emphasized its deployment of strong safety measures in response to concerns about misuse. The system blocks sexual deepfake creation while preventing watermark removal and rejecting CSAM requests. All generated images from OpenAI will possess standard C2PA metadata, which identifies them as OpenAI works despite lacking visual watermarks. The company operates internal tools that help verify images.
Shannon asserted that although systems cannot achieve perfection in this domain, they remain committed to refining their protective measures while considering this effort as the initial phase. Users who generate images through ChatGPT gain ownership rights to those images and can use them according to our specified usage policies.
OpenAI extends its flagship product capabilities through “Images in ChatGPT,” while advancing AI-driven creativity by providing users with a powerful visual expression tool in their conversational platform. Today’s launch represents a major advancement in AI tools through the combination of conversational AI technology and sophisticated image creation capabilities.
