Overview
- OpenAI rolled out ChatGPT Images 2.0, adding a ‘Thinking’ mode that can search the web, reason through prompts, and produce up to eight images from a single request.
- The update improves small text and icon rendering, handles dense layouts, supports wider aspect ratios, and outputs up to 2K resolution with stronger multilingual text.
- Images 2.0 is available across ChatGPT, with free users limited to a few daily images and paying tiers unlocking the Thinking workflows and higher throughput.
- Hands-on testing by ZDNET found it can create brand‑matched, publish‑ready graphics, yet it also made clear mistakes such as claiming nine items when only five were reviewed, swapping in an unreviewed vendor, and inventing star ratings.
- Experts warn the more realistic results raise the risk of deepfakes and non‑consensual sexual images, and designers online praise the power but note many outputs look samey, stoking worries about job impact and homogenized styles.