OpenAI Launches ChatGPT Images 2.0 with Reasoning Capabilities for Visual Workflows

OpenAI has released ChatGPT Images 2.0, positioning it as a visual workflow platform rather than just a creative tool. The model integrates reasoning capabilities when used with thinking or pro models, allowing analysis of complex tasks and generation of up to eight coherent outputs in a single request. Significant improvements include better instruction following, stronger text rendering, and enhanced object placement. Multilingual support has been strengthened, particularly for Japanese, Korean, Chinese, Hindi, and Bengali text rendering. The system handles various aspect ratios from 3:1 to 1:3, supporting formats like banners, slides, and social graphics. Images 2.0 shows improved fidelity across visual styles including photorealistic imagery, cinematic stills, pixel art, and manga. Available across ChatGPT, Codex, and the API, advanced reasoning outputs are limited to ChatGPT Plus, Pro, and Business users. OpenAI notes limitations in precise physical reasoning and highly detailed structural accuracy as areas for future development.