ChatGPT’s new Images 2.0 model is surprisingly good at generating text
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- OpenAI launched ChatGPT Images 2.0, which it "thinking capabilities" that let it search the web, generate multiple images from a single prompt, and double‑check output for accuracy.
- ChatGPT Images 2.0 can render non‑Latin scripts such as Japanese, Korean, Hindi, and Bengali with higher fidelity than previous models, according to OpenAI.
- Lesan AI founder Asmelash Teka Hadgu warned that diffusion‑based generators struggle with tiny text because they reconstruct images from noise, explaining past misspellings.
- DALL-E 3 from 2022 produced Mexican menu images riddled with nonsense words, highlighting the improvement seen in the new model.
- OpenAI declined to disclose the underlying model architecture powering Images 2.0 during a recent press briefing.
Why it matters: Marketers gain a tool that can instantly produce polished, multilingual visuals, cutting design turnaround time and reducing reliance on external agencies before Dec 2025, which should tighten campaign schedules and lower creative budgets.
