The new ChatGPT Images is here
ChatGPT Images is OpenAI’s image-generation and editing experience inside ChatGPT, powered by a new image model. It supports creating images from prompts, editing uploaded images, and exploring preset creative styles through a dedicated Images space.
What it does
ChatGPT Images can generate visuals from descriptions and make targeted changes to uploaded images. OpenAI says the model is designed to preserve important details such as lighting, composition, and people’s appearance while changing only the requested elements. Users can add, remove, combine, blend, or transpose image elements, apply stylistic transformations, and create more intricate compositions.
The service is also designed to handle text within images, including denser and smaller text. Preset filters and prompts in the dedicated Images area can help users explore styles and concepts without writing a prompt from scratch. The model is available in ChatGPT, while the corresponding API model is called GPT Image 1.5.
Notable capabilities
- Text-to-image generation from natural-language descriptions.
- Precise editing of uploaded photos, including adding, removing, combining, and blending elements.
- Creative transformations that preserve selected details while changing styles, layouts, or visual concepts.
- Improved instruction following for multi-element compositions and structured requests.
- Text rendering for graphics, posters, infographics, and other images containing written content.
Who it helps
The tool can support visual exploration, photo modifications, posters, advertisements, product concepts, social graphics, and other generated imagery. Marketing and ecommerce teams may also use GPT Image 1.5 through the API for branded visual work and product-image variations; OpenAI specifically describes logo preservation and catalog creation as API use cases.
How it fits a workflow
A typical workflow can begin with a written prompt or an uploaded image. Users can then refine the result through follow-up edits, asking the model to change particular subjects, styles, backgrounds, or layouts while preserving other elements. The dedicated Images space in ChatGPT provides preset ideas and filters for faster experimentation, and the feature is accessible from the sidebar on chatgpt.com and in the mobile app.
Developers can use GPT Image 1.5 through the OpenAI image-generation API or the Playground. OpenAI states that API image inputs and outputs are 20% cheaper than GPT Image 1.
Strengths and limits
The main strengths described by OpenAI are more reliable instruction following, better preservation during edits, improved rendering of text, and generation speeds of up to four times faster than the previous version. Results remain imperfect, however, with the publisher identifying limitations involving some styles, multiple faces, and multilingual image generation. Business and Enterprise access to the new ChatGPT Images experience was announced as coming later than the initial rollout.