OpenAI Images 2.0 arrives

OpenAI released ChatGPT Images 2.0 on April 21, adding a new image model to ChatGPT and the application programming interface, or API. (openai.com) The model is available in ChatGPT on all plans, while “images with thinking” is limited to paid plans when users pick Thinking or Pro models, according to OpenAI’s help documentation. Developers can access the same model in the API and in Codex under the name `gpt-image-2`. (help.openai.com) (community.openai.com) OpenAI said the upgrade improves text rendering, multilingual output, instruction following, layouts and editing, which targets a long-running weakness in image generators: garbled words inside signs, menus, slides and mockups. (openai.com) (community.openai.com) OpenAI’s system card says the new “thinking mode” can use reasoning and tools before generating an image, including live web search data and multiple image drafts from one prompt. That shifts the product from a picture maker toward a workflow tool for research, concept art and marketing comps. (deploymentsafety.openai.com) OpenAI’s launch materials show the model producing infographics, maps, magazine layouts, user-interface mockups and images with dense blocks of text. MacRumors reported the model can crawl the web for mockups and handle more detailed visual tasks than earlier versions. (openai.com) (macrumors.com) That wider range also sharpens an old risk. Business Insider reported OpenAI promoted the release with examples showing how realistic the model can make fake screenshots, fake magazine covers and other fabricated visuals. (businessinsider.com) OpenAI’s system card says the company tested the model for misuse including fraud, impersonation, misleading content and harmful instructions, and it says the product includes safeguards and policy enforcement. The same document says richer world knowledge, denser text and tool use can increase both utility and misuse potential. (deploymentsafety.openai.com) The release lands four months after OpenAI said its previous ChatGPT image upgrade, introduced in December 2025, delivered faster generation, more precise edits and more consistent details. Images 2.0 extends that push from prettier pictures into business documents and on-brand mockups that need legible words. (openai.com 1) (openai.com 2) For brand, design and communications teams, the immediate change is speed: one system can now draft a campaign visual, fill in readable copy, revise the layout and export more versions through ChatGPT or the API. The harder question is verification, because the same tools that make a polished mockup can also make a polished fake. (community.openai.com) (businessinsider.com)