GPT-4o Image Generator
GPT-4o image generation refers to the image capability OpenAI introduced on March 25, 2025. It supported images created and revised within a conversation, with uploaded images as references. This page preserves that historical name. OpenAI’s original announcement now points readers to the newer ChatGPT Images experience.
What it offers
The 2025 release focused on rendering text, following detailed instructions, and revising a subject through multiple turns. Those capabilities made posters, illustrations, and visual drafts practical use cases. They did not guarantee correct lettering, anatomical detail, or exact preservation of an uploaded subject. Current image models have their own capabilities and limitations, so old comparisons do not establish today’s performance.
Access and cost
GPT-4o’s text-model retirement in ChatGPT on February 13, 2026 did not itself retire ChatGPT Images. OpenAI explicitly distinguishes the image model from the text model. For new work, use the current image interface in ChatGPT or the documented GPT Image API. The current API guide recommends GPT Image 2.5 models; the historical name gpt-4o is not a substitute for an image-generation model identifier.
Practical limits
ChatGPT plans set their own image access and limits. API requests are billed separately for the selected model, with costs affected by inputs, image size, and quality. When moving an old workflow, review request parameters and result handling and compare example edits. Preserve source images and prompts. Do not promise unlimited free images or identical results from a newer model.