What is new in ChatGPT Images? OpenAI rolls out GPT Image 1.5 with faster generation and sharper edits

OpenAI on December 17 announced a major upgrade to ChatGPT Images with the release of GPT Image 1.5, its flagship image generation model, significantly enhancing how users create and edit images using artificial intelligence. The update is being rolled out globally to all ChatGPT users and API customers, with the company stating that it delivers faster image generation, more precise edits and improved preservation of original image details.

According to OpenAI, the latest ChatGPT Images model enables users to generate images from scratch or edit existing photos with substantially higher accuracy, producing images up to four times faster while maintaining consistency in key elements such as lighting, composition and people’s appearance across edits. The update also introduces a new Images feature within ChatGPT that offers a dedicated creative space with preset styles, filters and prompts to make image exploration and experimentation easier.

A key improvement lies in instruction-following, with OpenAI stating that when users edit an uploaded image, the model now modifies only the elements specified in the prompt while preserving everything else. This allows for more realistic photo edits, believable clothing and hairstyle try-ons, and cleaner stylistic filters and conceptual transformations. The model supports advanced editing functions including adding, removing, blending, combining and transposing elements without degrading the original image quality.

OpenAI further stated that GPT Image 1.5 delivers stronger creative capabilities, allowing users to add text, adjust layouts and reimagine visuals while retaining the essence of the original image. Text rendering has been improved to handle denser and smaller text more accurately, alongside quality upgrades such as better handling of multiple small faces and more natural-looking outputs suited to real-world applications.

The same enhancements are available through the GPT Image 1.5 API, which OpenAI said offers significantly better image preservation and brand consistency than its predecessor. This makes the model particularly useful for marketing and brand design, logo and graphic creation, and ecommerce product catalogues generated from a single source image. OpenAI also informed that it has reduced image input and output costs by 20 per cent compared with GPT Image 1, enabling developers and businesses to generate more images within the same budget.

Developers can access the model through the OpenAI Playground and refer to the official prompt guide for best practices, the company stated. The new ChatGPT Images model is available globally across ChatGPT and the API with no manual model selection required, while the earlier version of ChatGPT Images will continue to remain accessible as a custom GPT.