OpenAI launched GPT Image 2 with thinking capabilities, web search integration, and batch generation of up to 8 consistent images from one prompt.
OpenAI announced ChatGPT Images 2.0, powered by the new GPT Image 2 model, with 'thinking capabilities' available to Plus, Pro, Business, and Enterprise subscribers. The update enables the image generator to pull real-time web data, create up to 8 consistent images from a single prompt, and generate visual explainers from uploaded files. The model also improves text rendering across Latin and non-Latin scripts including Japanese, Korean, Chinese, Hindi, and Bengali. Additional improvements include better instruction-following, detail preservation, and support for aspect ratios ranging from 3:1 to 1:3.
GPT Image 2 introduces real-time web retrieval before image generation — a meaningful architectural shift from static prompt-to-image pipelines. The batch generation of up to 8 consistent images from a single prompt could replace multi-call workflows and reduce orchestration overhead for apps building storyboards, product catalogs, or sequential visual content. The model isn't API-available at launch for all tiers, so direct integration may lag the ChatGPT UI rollout.
If you're running an image generation pipeline for consistent character or product visuals, test the 8-image batch endpoint this week — measure whether a single GPT Image 2 call replaces your current multi-step inpainting loop.
Open chat.openai.com and select a thinking-enabled model (Plus or Pro required)
Tags
Related