ChatGPT just launched Images 2.0, and it finally fixes warped text

OpenAI has unveiled ChatGPT Images 2.0, a major overhaul to its image-generation capabilities that could make AI visuals far more useful for designers, marketers and everyday users.

While earlier AI image tools often generated results with frustrating flaws such as warped text, poor layouts and vague prompt-following, OpenAI promises Images 2.0 is different.

According to OpenAI, the new model delivers stronger instruction following, sharper text rendering, improved multilingual support and more control over composition, aspect ratio and visual consistency. In short: less “AI art experiment,” more “usable design tool.”

Why this upgrade matters

For many designers, AI image tools have been hard to trust for real work. Now, users can potentially create presentation slides, social media graphics, banners, posters and product mockups directly inside ChatGPT.

OpenAI says Images 2.0 changes that by improving the exact pain points creative pros complain about most:

• Better small text and typography
• More accurate object placement
• Cleaner layouts with whitespace and hierarchy
• Stronger handling of posters, explainers and UI elements
• More realistic styles and sharper consistency
• Support for aspect ratios as wide as 3:1 and tall as 1:3

Designers may actually care this time

The biggest shift may be intent. Instead of focusing only on surreal art or novelty images, OpenAI is positioning Images 2.0 as a strategic design system, which is something that can help carry a project from rough idea to finished asset.

The company says the model can reason through layouts, use web information when a thinking model is selected and even generate up to eight related images at once with continuity across characters or objects.

That could be valuable for ad campaign variations, storyboards, social media assets, comic sequences, product launches and multi-language marketing materials.

In fact, one of the most notable improvements is support for non-English text rendering. OpenAI says Images 2.0 makes significant gains in Japanese, Korean, Chinese, Hindi and Bengali, helping generate visuals where language is part of the design rather than an afterthought.

Bottom line

OpenAI says ChatGPT Images 2.0 is rolling out now to all ChatGPT and Codex users. Advanced outputs powered by thinking models are reserved for Plus, Pro, Business and Enterprise users. The underlying gpt-image-2 model is also available through the API for developers.

For the first time, this ChatGPT image upgrade feels as if its less about images going viral and more about getting actual work done. Users might find that Images 2.0 is the version that legitimatly supports creative workflow.

Follow Tom's Guide on Google News and add us as a preferred source to get our up-to-date news, analysis, and reviews in your feeds.

More from Tom's Guide

• I built a pizza startup brand in 30 minutes with Claude Design — it looked launch-ready
• Reese Witherspoon’s AI push is sparking a firestorm — but she’s actually right
• I asked ChatGPT to use Elon Musk’s ‘Relevance Rule’ to fix my memory — and I’m never going back to notes