Google 模型
Google Nano Banana, Nano Banana 2, 2 Lite, 2.1 and Pro via API: Gemini-based text-to-image and image editing models from Google DeepMind.
Google develops the Nano Banana text-to-image and image editing models through Google DeepMind, the AI unit it formed in April 2023 from DeepMind and Google Brain. The company was founded in 1998 by Larry Page and Sergey Brin, is headquartered in Mountain View, California, and has been part of Alphabet since 2015.
可用模型
google/nano-banana-2.1/text-to-image文生图按 token · 默认约 $0.04
google/nano-banana-2.1/image-to-image图生图按 token · 默认约 $0.042
Nano Banana 2
google/nano-banana-2/text-to-image文生图$0.08 / 张
google/nano-banana-2/image-to-image图生图$0.08 / 张
google/nano-banana-2-lite/text-to-image文生图按 token · 默认约 $0.042
google/nano-banana-2-lite/image-to-image图生图约 $0.042 / 张
Nano Banana
google/nano-banana/text-to-image文生图约 $0.04 / 张
google/nano-banana/image-to-image图生图约 $0.04 / 张
google/nano-banana-pro/text-to-image文生图$0.15 / 张
google/nano-banana-pro/image-to-image图生图$0.15 / 张
关于厂商
Google builds the Gemini AI models, and Nano Banana is the Gemini line for image generation and editing. Each version turns a text prompt, with optional input images, into a new or edited image; newer versions take several reference images and render legible text inside images.
- Nano Banana: Text-to-image and image editing on Gemini 2.5 Flash Image, released in August 2025; keeps characters consistent, with up to three input images recommended.
- Nano Banana 2: Text-to-image and image editing on Gemini 3.1 Flash Image, released in February 2026; supports Google Search grounding, text translation and 512px to 4K output.
- Nano Banana 2 Lite: Text-to-image and image editing on Gemini 3.1 Flash-Lite Image, the Lite tier of Nano Banana 2; outputs 1K (1024px) images only.
- Nano Banana 2.1: Text-to-image and image editing as an update to Nano Banana 2; supports Google Search grounding and adjustable thinking levels.
- Nano Banana Pro: Text-to-image and image editing on Gemini 3 Pro Image, released in November 2025; renders detailed text in images and outputs up to 4K.
通过 API 接入
一把 MuseSpan API 密钥即可调用本页所有 Google 模型：复制模型 ID，填进请求的 model 字段。