OpenAI is gearing up to launch its next generation of image generation models alongside the anticipated GPT-5.2. The new models, referred to as Chestnut and Huzzlenut in testing environments, will be branded as Image-2 and Image-2-mini. They aim to improve upon the previous Image-1, focusing on enhancing visual quality and versatility. Early feedback from testers indicates a significant boost in image detail and fidelity, addressing issues like the yellow tint that plagued Image-1 outputs. These advancements bring OpenAI's technology closer to Google’s Nano Banana 2, especially regarding detail and color accuracy. While Nano Banana 2 still excels in some areas, OpenAI has narrowed the gap, particularly by resolving past color issues and improving overall image structure. The visual style of the new models seems to echo earlier DALL-E versions, combining established strengths with upgraded rendering capabilities. Although there’s no confirmed launch date, clues in ChatGPT's code suggest an announcement is imminent. The expectation is that the release will coincide with GPT-5.2. This strategy aligns with OpenAI's approach of making rapid, incremental enhancements to both text and visual AI, aiming to compete effectively with Google’s evolving multi-modal offerings. If launched as planned, these new models will benefit creative professionals, developers, and businesses by delivering higher-quality assets for design, marketing, and prototyping tasks.