OpenAI announced ChatGPT Images 2 on Tuesday, a new image model designed specifically to produce text-heavy images such as infographics, scientific posters, study guides and marketing materials; the company says the model is rolling out to all users now.
The announcement comes with a public pitch about who the feature is for. Adele Li, speaking about the company’s wider vision for ChatGPT, said: "The aperture and use cases for visual intelligence just expand so broadly, and we believe that this is so critical to ChatGPT's vision for developing your own personal assistant, because your creative assistant is a huge part of who you are as an individual." OpenAI also framed the release as part of a push toward "economically valuable creative tasks."
ChatGPT Images 2 follows a rapid cadence: OpenAI released ChatGPT Images 1.5 four months earlier and shuttered its Sora AI video app one month before this announcement. The company says Images 2 improves typography, iconography and composition to produce more professional AI images, and that it is now the company's best model for rendering legible, factually correct text.
The practical improvements are explicit. OpenAI says Images 2 can generate text in multiple languages and is tuned for assets people actually publish: infographics, posters, study guides and marketing materials. The rollout is happening now for all users, but how much anyone can generate will depend on their plan: generation limits vary by subscription. Paying users also gain the ability to create images using the company’s thinking and reasoning models.
For developers, OpenAI opened the model through its API with higher-resolution options. Teams integrating Images 2 can generate images in 2K and 4K, though those API resolutions remain in beta for now. That mix — broadly available consumer access alongside beta developer tools — is meant to let creators test professional outputs while the company stabilizes the highest-resolution pipelines.
Context matters here. OpenAI has been building toward a super app around its Codex platform, and ChatGPT Images 2 is positioned as the creative piece of that plan. Historically, AI image systems have struggled to render legible and factually correct embedded text; OpenAI frames this release as a technical answer to that problem. Competitors have been moving on the same front: Google improved text rendering with Nano Banana Pro before this article was published, sharpening the comparison between offerings.
The announcement also exposes a tension in OpenAI’s strategy. The company closed its Sora video app just one month before pushing Images 2, suggesting a reallocation of resources toward still-image visual intelligence rather than short-form video. At the same time, key capabilities remain gated: the highest-resolution outputs are in beta, and generation limits differ by plan. That creates a two-track user experience — broadly available image generation on one hand and a paid or developer-heavy route to premium outputs on the other.
Those frictions point to what comes next. If Images 2 truly delivers legible, multi-language text inside polished visuals, it would remove a primary barrier that kept designers and educators from adopting AI-generated images for publication. But the product’s impact will depend on rollout details: whether beta resolutions scale, how strict plan-based limits are, and how quickly OpenAI stitches these image capabilities into ChatGPT’s wider assistant features. The company has signaled its aim plainly: to enable "economically valuable creative tasks."
In the end, OpenAI is betting that more fluent visual intelligence will make ChatGPT a more central creative tool. Adele Li’s framing — that a creative assistant is core to personal identity — is the thesis behind this release. If Images 2 can make professional, text-rich graphics reliable and accessible, the company will have moved a long way toward turning gpt-powered image generation into everyday creative work.
