ChatGPT latest update: Finally gets text and object placement right in images

OpenAI launches ChatGPT Images 2.0 with improved accuracy, text rendering, flexible formats, and multi image outputs for real world creative tasks.

OpenAI has introduced ChatGPT Images 2.0, a big upgrade to its image generation model. It can create much more real images with accurate text with multiple languages support. The new model supports up to 8 coherent images with one request. It is rolling out to ChatGPT, Codex and its developer API.

ChatGPT Images 2.0 brings sharper, more practical AI image creation. (OpenAI)

By Amit Rahi

For the past seven years, I have tracked consumer tech through constant shifts in hardware, platforms, and the way people actually use devices. Covering everything from budget gear to flagship hardware, I focus on what readers need to know, not on buzzwords or launch cycle hype. My expertise spans gaming laptops and chairs, high-performance PCs, gaming monitors, printers, smartwatches, earphones, headphones, Bluetooth speakers, tablets, and more, with a particular emphasis on how these products hold up in daily use. Reviews, explainers, buying guides, and news pieces all share the same goal: giving readers enough detail to make confident decisions without wading through fluff. Away from deadlines, I spend a lot of time gaming and watching films and anime, which naturally filters back into the work. Performance, comfort, display quality, and sound are judged the way players and viewers experience them, not just by lab numbers, which keeps my coverage grounded in real scenarios rather than just benchmarks.

Read moreRead less

According to OpenAI, the Images 2.0 significantly improves how accurately the system follows detailed prompts. It can now create images with dense text more clearly, it can position objects with better precision and now supports more layouts than before. Professionals can now use ChatGPT to create UI mock-ups, diagrams, presentations and other marketing assets.

This update introduces “thinking enabled” workflows within ChatGPT. This allows it to reason through complex prompts, pull relevant context when needed, and generate multiple distinct outputs from a single request. Users can now create up to 8 images at once making it easier to create storyboards, comics panels or presentation assets.

One of the key additions is expanded aspect ratio support, ranging from 3:1 to 1:3. This allows users to generate images tailored for different formats, including banners, posters, slides, and mobile screens, without needing external adjustments.

For developers, the new model is available via the API as gpt-image-2, enabling direct integration of image generation and editing into apps and workflows.

ChatGPT Images 2.0 is available starting today for all ChatGPT and Codex users. Advanced outputs powered by the thinking system are limited to paid tiers, including Plus, Pro, Business, and Enterprise plans.