OpenAI Ships Images 2 with Instant & Thinking Modes
GPT-Image-2 drops with 2K resolution, native text rendering, and two speed tiers — live on ChatGPT and API now.
OpenAI just launched ChatGPT Images 2 (GPT-Image-2), its most capable image generation model yet, available in two flavors: Instant (fast) and Thinking (higher-quality, reasoning-first).
The upgrades are meaningful. Text rendering inside images — long a glaring weakness — is now solid, including non-Latin language support. Resolution hits 2K, and the model can pull current info from the web before generating, giving it an effective knowledge cutoff of December 2025. OpenAI's demo framing: it can handle entire magazine layouts end-to-end, from copy to design.
Both versions are live in ChatGPT and the API today. Separately, OpenAI also announced GPT-5.2 in Instant, Thinking, and Pro tiers — rolling out to paid users with improvements in vision, long-context, and agentic tool use.
Why it matters: two-tier image generation (speed vs. quality) mirrors how reasoning models are sold — and signals OpenAI is treating image gen as a first-class product, not a side feature.
Sources
Written by an AI pipeline from the sources above. How it works.
Feed, daily deep-dive and bytes — readable offline, with push alerts for the topics you follow.