OpenAI launches ChatGPT Images 2.0 with improved text rendering

OpenAI has launched ChatGPT Images 2.0, a model that significantly improves the rendering of text and fine-grained elements like iconography and user interface (UI) components. Historically, image generators struggled with spelling, often producing illegible text. The new model addresses this by effectively generating text within images, such as creating a restaurant menu that lists ceviche priced at $13.50 without the spelling errors typical of older models like DALL-E 3. Additionally, the model shows marked progress in handling non-Latin text, specifically in Japanese, Korean, Hindi, and Bengali.

In a press release, OpenAI stated: “Images 2.0 brings an unprecedented level of specificity and fidelity to image creation. It can not only conceptualize more sophisticated images, but it actually brings that vision to life effectively, able to follow instructions, preserve requested details, and render the fine-grained elements that often break image models: small text, iconography, UI elements, dense compositions, and subtle stylistic constraints, all at up to 2K resolution”. The model introduces what OpenAI calls “thinking capabilities,” which the company claims allow the system to search the web, generate multiple images from a single prompt, and double-check its own creations. However, users should note that the model’s knowledge cutoff—the date after which the model has no training data—is December 2025.

Historically, AI image generators have struggled to spell because they generally used diffusion models, which are AI models that reconstruct images from noise. “The diffusion models […] are reconstructing a given input,” Asmelash Teka Hadgu, founder and CEO of Lesan AI, told TechCrunch in 2024. “We can assume writings on an image are a very, very tiny part, so the image generator learns the patterns that cover more of these pixels.” While researchers have explored other mechanisms like autoregressive models, OpenAI declined to answer what kind of model powers ChatGPT Images 2.0.

Access to the new model begins Tuesday for all ChatGPT and Codex users, with paid tiers offering more advanced outputs. OpenAI is also releasing the gpt-image-2 API for developers, with pricing dependent on the quality and resolution of outputs.

Why it matters

The release of ChatGPT Images 2.0 marks a significant improvement in AI image generation, specifically in the ability to render text and fine-grained elements like iconography and UI elements, which have historically been difficult for diffusion-based models.