OpenAI's Images 2 hits Azure

OpenAI’s new image model is landing inside Microsoft’s Azure AI Foundry, giving enterprise teams a way to generate campaign visuals without leaving Microsoft’s development stack. (techcommunity.microsoft.com) OpenAI added ChatGPT Images 2.0 to ChatGPT on April 21, 2026, and said the model is available across all ChatGPT plans. The company described it as its new image-generation model and paired the launch with a new “images with thinking” feature. (help.openai.com) Microsoft said April 21 that OpenAI’s GPT-image-2 is now generally available and “rolling out today” in Microsoft Foundry. The company said the model adds multilingual understanding, improved instruction following, higher-resolution support, and a routing layer for production workflows. (techcommunity.microsoft.com) Image generation models turn text prompts into pictures, and the business pitch is speed: one team can create many versions of the same ad without a new photo shoot. Microsoft’s example was a small design team localizing a global social campaign for multiple markets. (techcommunity.microsoft.com) That puts the release in the middle of a broader shift from image creation as a consumer demo to image creation as office software. Azure AI Foundry is Microsoft’s platform for building, evaluating, deploying, and governing artificial intelligence systems in one workflow. (techcommunity.microsoft.com) OpenAI has been pushing the same direction in its own products. Its developer documentation says GPT Image models support multi-turn editing, image inputs, and generation inside the Responses API, which lets developers fold image work into larger app flows instead of treating it as a separate tool. (developers.openai.com) The consumer and enterprise naming is slightly different. OpenAI’s help center calls the launch “ChatGPT Images 2.0” inside ChatGPT, while Microsoft markets the Azure version as “GPT-image-2” in Foundry. (help.openai.com; techcommunity.microsoft.com) Early coverage of the launch focused on one old weakness of image models: text inside images. 9to5Mac reported that OpenAI’s preview materials pointed to stronger text rendering, which would make the model more useful for posters, mockups, and ad creative that need readable words, not just striking pictures. (9to5mac.com) The immediate test is not whether the model can make a single polished image, but whether teams can produce dozens of on-brand versions fast enough to replace parts of a campaign workflow. Microsoft is betting Azure customers want that work inside Foundry, not passed between separate chat, design, and production tools. (techcommunity.microsoft.com)