OpenAI has rolled out Images 2.0, a new image‑generation engine inside ChatGPT that promises sharper detail and tighter adherence to prompts. The update builds on last year’s launch, adding reasoning skills that let the model browse and verify content. Early tests show it handles dense text and complex scenes with unprecedented clarity for designers and developers.

Images 2.0 also delivers marked improvements in rendering non‑Latin scripts, tackling Japanese, Korean, Chinese, Hindi and Bengali with higher fidelity. The model faithfully reproduces stylistic nuances of different visual languages, making it attractive for game prototyping and storyboard creation. Users can now choose aspect ratios up to 3:1 or 1:3 and output resolutions up to 2K.

Beyond language, the engine supports up to eight simultaneous outputs and can generate images up to 2K in a single pass. A preview test produced a pixel‑art portrait of a tortoiseshell cat in the style of Pokémon’s third‑generation titles, followed by a transparent PNG conversion and a four‑page manga—all within minutes for creative teams.

Images 2.0 is live for all ChatGPT tiers, with Plus and Pro subscribers accessing richer outputs. The model is also exposed via the OpenAI API and Codex app, extending its reach to developers and designers. This release follows Anthropic’s entry into visual design, tightening competition for both creators and studios today.