- calendar_today August 10, 2025
OpenAI introduced “Images in ChatGPT,” which brings direct image generation capabilities into the ChatGPT platform. The recently released GPT-4o model powers this advancement which allows users to generate images during chat interactions and represents a substantial improvement in AI-generated content.
“Images in ChatGPT” delivers powerful image generation tools to everyone through all ChatGPT subscription levels, including Plus, Pro, Team, and free access. OpenAI spokesperson Taya Christianson confirmed that free tier users face similar usage limits as DALL-E 3 at three images per day, although these limits can change according to demand. Users who like DALL-E can continue to use it through a custom GPT interface.
OpenAI’s research head Gabriel Goh described GPT-4o as an “omnimodal” foundation that processes text, image, audio, and video data types. The model now features improved “binding” ability, which solves frequent issues faced during AI image generation. GPT-4o stands out from earlier models by maintaining clear object-attribute relationships for 15 to 20 objects without any shape or color confusion.
A major advancement in the system is its exceptional ability to render text. AI-generated images traditionally display garbled or nonsensical text. Goh characterized the development as an extensive iterative process that extended over several months until completion. Although perfect text rendering for small elements remains unattainable, the team has reached sufficient consistency to make image text reliably usable.
Unlike the diffusion models that most image generators use, this system’s architecture employs an autoregressive approach. The sequential image generation process from left to right and top to bottom mirrors text production methods, which researchers believe enhances text rendering and binding functions.
OpenAI exhibited their system’s various uses at a briefing, which demonstrated scientific diagram generation like Newton’s prism experiment with precise labeling and the creation of multi-panel comics with consistent characters and dialogue, along with informational poster design featuring accurate text. The demonstration included practical use cases where the system generated transparent background images for stickers, restaurant menus, and logos.
ChatGPT’s multimodal product lead, Jackie Shannon, highlighted how the system utilizes global knowledge. When drawing an image, she relies on her personal skillset while also applying her acquired global knowledge. The model incorporates world knowledge into its operations, which enables you to request an image of Newton’s prism experiment without needing to provide background information about it.
OpenAI believes that the improved quality and enhanced features of image generation are worth the longer processing times. Shannon acknowledged the need for latency improvements but emphasized that the high quality and advanced capabilities of these images compensate for the additional wait time.
Key Technological Advancements: Binding, Text Rendering, and Architectural Shifts
GPT-4o delivers major technological improvements through its “binding” capabilities, which enable precise visualization of intricate scenes containing multiple objects. The improved text rendering solution, which emerged from extensive iterative development, resolves a key shortcoming found in earlier AI image generators. The move towards autoregressive image generation methods instead of traditional diffusion models is believed to have played a significant role in these technological improvements.
Safeguards and User Empowerment: Addressing Misuse and Ensuring Responsible AI
OpenAI responded to potential misuse worries by emphasizing its strong protective measures. The system incorporates preventative measures against the removal of watermarks and also blocks the creation of sexual deepfakes while rejecting CSAM requests. While visual watermarks are not present on generated images, each image includes standard C2PA metadata that identifies them as creations of OpenAI. The company operates internal tools dedicated to image verification.
Shannon pointed out that although every system shows imperfections in this field, we constantly work on enhancing our protective measures and regard this as our initial step. Users who generate images through ChatGPT become the owners of these images and can freely use them according to our usage policies.
OpenAI is boosting ChatGPT functionality while positioning itself as a leader in user-friendly AI image creation technology and simultaneously working to mitigate this tech’s inherent risks.
