ChatGPT’s New Image Model Fixes Photos, Adds Objects, and Follows Instructions Better, Prompts Inside

OpenAI has introduced a new version of its image‑generation model for ChatGPT, bringing noticeable improvements in speed and accuracy. The update, which is rolling out to all users, aims to make the tool more dependable for everyday image tasks, rather than just creative experiments.

 A More Controlled Editing Experience

The latest model focuses on solving a long‑standing issue: unpredictable edits. Earlier versions often changed parts of an image that users didn’t intend to modify, especially when making small adjustments. The new system is designed to follow instructions more closely, keeping faces, lighting, and backgrounds consistent while applying only the requested changes.

The model also produces images faster than before, which may help users who rely on multiple iterations to refine a design or correct a photo. Text rendering has been improved as well, making posters, labels, and social media graphics clearer and easier to read.

 A Dedicated Space for Image Creation

ChatGPT now includes a separate “Images” section that groups together preset styles, trending prompts, and tools for reusing a user’s likeness across different creations. The interface allows new images to be generated while others are still processing, making it easier to test variations without waiting for each one to finish.

The update is likely to appeal to a broad range of users. Students can create diagrams or simple posters for school projects. Small businesses can produce basic marketing visuals without relying on external design tools. Content creators can generate thumbnails or edit photos for social media. Casual users may find it useful for cleaning up personal photos or making quick greeting cards. The update does not change how users interact with ChatGPT, but it does make image generation more predictable and practical. 

 Prompts You Can Try 

To highlight the improvements, here are examples of prompts that take advantage of the new model’s precision and speed:

- “Remove the crowd in the background but keep the lighting the same.”

- “Add a small dog sitting next to me without changing the rest of the photo.”

- “Create a simple birthday card with pastel colours and handwritten‑style text.”

- “Turn this picture into a comic‑style illustration while keeping the face accurate.”

- “Make a clean product photo on a white background with soft shadows.”

- “Add bold text at the top that says ‘Winter Sale’ in a modern font.”

Read More: Nano Banana AI Saree Trend is Everywhere, Here's How to Create Your Own Viral Look Using These Simple Prompts