ChatGPT Images 2 draws multi-image examples

OpenAI’s image model is drawing a new round of user demos centered on one specific claim: it can hold together more than one image at a time. Over the past 48 hours, X users shared side-by-side outputs they said were made with ChatGPT Images 2.0, including multi-image prompt tests, recurring-character examples and software-style interface mockups. The posts landed about a month after OpenAI introduced ChatGPT Images 2.0 on April 21 and said the model improved text rendering, multilingual support and visual reasoning. ### Which posts are people pointing to? Two May 22 X posts became representative examples in the latest burst of sharing. One post by VengeonsP showed multi-image outputs and described the model’s ability to keep visual details aligned across frames, according to the post referenced in the source briefing. Another post by alukyi shared UI-style mockups generated inside a Codex workspace, also on May 22, according to the briefing and linked X post. (openai.com) The examples match the kinds of use cases circulating more broadly in recent social discussion. The social briefing for this story described “heavy buzz” around ChatGPT Images 2.0, with users praising realism, consistency in multi-image prompts and workflow use inside Codex, while also noting some queuing and pinning bugs. That briefing also cited filmmaking, scrapbook aesthetics and character consistency as recurring themes in the posts. (community.openai.com) ### What did OpenAI actually launch in April? OpenAI said on April 21 that ChatGPT Images 2.0 introduced a “state-of-the-art image generation model” with improved text rendering, multilingual support and advanced visual reasoning. In a separate developer announcement the same day, OpenAI said gpt-image-2 was available in the API and in Codex. OpenAI’s developer post described the model as built for “production workflows,” saying it offered stronger editing, better layouts, improved text rendering and more reliable instruction-following. (community.openai.com) The company also said the model could create assets for apps, ads, presentations and documents, and highlighted more practical text-heavy visuals, structured generation and better composition control. (openai.com) ### Why are multi-image and UI examples getting attention? OpenAI’s own materials point to the same capabilities users are testing. The company’s prompting guide for gpt-image-2 says the model supports robust facial and identity preservation for edits, character consistency and multi-step workflows, along with reliable text rendering and multi-panel compositions. (community.openai.com) Codex is part of that story because OpenAI said developers can use gpt-image-2 “in Codex when you want to create visual assets directly from what you are already building.” That makes UI mockups a natural demo category: users are showing not only standalone pictures, but images tied to software and product workflows. (developers.openai.com) ### Where is the product available now? OpenAI’s Help Center says ChatGPT Images 2.0 is available on all ChatGPT tiers and can be used on web, iOS and Android. The same help page says “Images with thinking” is available on Plus, Pro and Business, and is coming to Enterprise and Edu soon. The Help Center also says users can create an image in a conversation, edit generated or uploaded images, choose aspect ratios and save outputs under the Images section for reuse. (community.openai.com) OpenAI’s API documentation separately lists GPT Image 2 as its state-of-the-art image generation model for high-quality generation and editing. ### What comes next for these examples? (help.openai.com) May 23 is likely to bring more public testing because the model is already live in ChatGPT, the API and Codex, according to OpenAI’s April 21 announcements. The next visible proof points are likely to remain on X and in OpenAI’s own docs, where the company is publishing prompting guides and product updates for gpt-image-2 and ChatGPT Images 2.0. (openai.com) (help.openai.com)