GPT Image 2 is OpenAI's upcoming next-generation image model. Preview the announcement, browse features, and generate with a comparable model today while GPT Image 2 is getting ready.
このページはアナウンスです。モデルはまだ利用できません。現在このプラットフォームでの生成は、同等の能力を持つ別モデルが担っており、本サイトは上流モデルの提供元と公式な関係はありません。
Early demos suggest GPT Image 2 renders real brands, interfaces, and locations with striking accuracy, pulling on GPT's world knowledge.
Readable, scene-integrated text — handwritten notes, speech bubbles, signage, posters — in a way previous diffusion models struggled with.
Early reports say GPT Image 2 has dropped the warm yellow cast that became a signature flaw of GPT-Image-1.
GPT Image 2 is expected to compose complex scenes from long prompts without dropping details — the kind of multi-clause descriptions that used to require manual touch-ups. Until release, these previews are produced with Nano Banana Pro, which is closest in capability today.
プロンプト
Editorial photo: a chef plating a seven-course tasting menu at a stainless steel pass, low-key lighting, dramatic steam, legible handwritten tickets pinned behind
Preview (Nano Banana Pro)

Preview (Nano Banana Pro)

Text that belongs in the scene — signage, speech bubbles, handwritten notes — rendered with correct letterforms and proper placement. This has historically been the hardest single capability for diffusion image models.
プロンプト
Subway poster for a fictional film titled 'Night Shift', bold grotesque sans-serif, tagline 'The city never sleeps. Neither did we.', clean minimal layout
Preview (Nano Banana Pro)

Preview (Nano Banana Pro)

Reflections, shadows, and material response that read as physically plausible. Golden hour, neon environments, glass refraction — the edge cases that expose older models.
プロンプト
Vintage motorcycle parked in an empty parking garage lit only by a single sodium vapor lamp, long orange shadows, wet concrete reflections
Preview (Nano Banana Pro)

Write a concrete, detail-rich prompt. Coming-soon features aside, this is the biggest lever for image quality.
Your prompt opens in Workspace. While GPT Image 2 is unreleased, generation routes to Nano Banana Pro — you'll see a banner confirming the substitution.
Choose 1K for thumbnails, 2K for hero use, 4K for print or zoomable detail.
Refine the prompt, swap references, tweak aspect ratio. When GPT Image 2 launches, your flow stays the same — we just flip the backend.
GPT Image 2 is not yet available. Tap below to open Workspace and generate with Nano Banana Pro in the meantime — we'll flip the backend the moment OpenAI releases the model.
Try in Workspace
Nano Banana で美しい AI 画像を生成。高速、多機能、無料で開始できます。テキストプロンプトで美しいアートワーク、写真、デザインを作成しましょう。

Nano Banana Pro でプロフェッショナルな AI 画像生成。より高い解像度、優れたディテール、プロクリエイター向けの高度な機能。

最新の Nano Banana 2 モデル。より手頃な価格で、最大 14 枚の参照画像をサポートし、低コストで優れた品質を実現。

Z Image で高速かつ手頃な AI 画像生成。テキストのみの生成で 1 画像あたりわずか 1 credit。素早いイテレーションに最適。