Overview
- ChatGPT Images 2.0, which OpenAI rolled out Tuesday, reaches all ChatGPT and Codex users, with advanced thinking features limited to Plus, Pro and Business tiers.
- When a thinking model is selected, the system can search the web, plan layouts, verify results, and generate up to eight coherent images from a single prompt.
- The update strengthens dense text rendering and non‑Latin scripts such as Japanese, Korean, Chinese, Hindi and Bengali, and adds 2K output with aspect ratios from 3:1 to 1:3.
- Developers can use the model via the API under the gpt-image-2 identifier and inside the Mac Codex app for workflows like UI mockups, comics and multi-image design sets.
- OpenAI and reviewers still note gaps with physical-world tasks and ultra-fine, repetitive textures, keeping human review important as Google, Anthropic and Microsoft intensify their visual AI offerings.