OpenAI launches ChatGPT Images 2.0 with improved text rendering and non-Latin script support

OpenAI has released ChatGPT Images 2.0, a major upgrade to its image generation model that significantly improves text rendering capabilities and supports non-Latin scripts. The new model is available globally to ChatGPT and Codex users, with a more powerful version for paying subscribers. The update introduces "thinking capabilities" that allow the model to search the web for recent information before generating images. This enables creation of marketing assets with up-to-date details and multi-panel comic strips. Images 2.0 can generate multiple images from a single prompt and produce outputs up to 2K resolution. A key improvement is the model's ability to render text in languages like Japanese, Korean, Hindi, Bengali, and Chinese. Previous image generation models often struggled with accurate text rendering, particularly for non-Latin scripts. The new version also offers more customizable aspect ratios, ranging from 3:1 wide to 1:3 tall. OpenAI is offering two versions of the model: Instant for faster generation and Thinking for more complex tasks that require research and multi-step reasoning. The company says the model can create entire magazines and complete design projects with text-heavy layouts. The knowledge cutoff for Images 2.0 is December 2025, making it more capable of generating images related to recent events and information. Complex image generation takes just minutes compared to longer processing times with previous models. Alongside the image model update, OpenAI announced Codex Labs, a new enterprise program to help organizations integrate its coding assistant into workflows. The initiative brings OpenAI experts directly into companies for hands-on workshops and deployment guidance. The gpt-image-2 API is now available with pricing dependent on output quality and resolution. Users can access the new capabilities immediately through ChatGPT and Codex interfaces.