OpenAI ships GPT-Image-2 with reasoning, 4K output, and near-perfect text rendering
What Happened
OpenAI released GPT-Image-2 on April 21, the successor to DALL-E 3 and Image 1.5. New capabilities include native reasoning, 4K resolution, near-perfect text rendering, and stronger support for Japanese, Korean, Chinese, Hindi, and Bengali. The model is available to all ChatGPT and Codex users, with thinking mode reserved for paid tiers and the API.
My Take
Coverage focused on the text-rendering jump. The deeper change is reasoning inside an image model, which collapses the line between "generate a picture" and "do a small design job." Posters, slides, infographics, mocked-up dashboards — work that used to require a human designer with Figma now becomes a single prompt. Prediction: within six months, at least one major design tool drops its generative bolt-on and ships a thin wrapper around GPT-Image-2 instead, conceding the model layer to OpenAI.
Read Original Source