谷歌推出Nano Banana 2 Lite（Gemini 3.1 Flash Lite Image），成为最快且最便宜的Gemini图像模型
英文摘要
Google has introduced Nano Banana 2 Lite, officially named Gemini 3.1 Flash Lite Image (API identifier gemini-3.1-flash-lite-image), marketed as the fastest and most affordable model in the Gemini image generation family. The model is designed for high-velocity, large-scale use. In a quick test using Google AI Studio, Simon Willison generated a "Where's Waldo"-style image featuring a raccoon with a ham radio, and found the result more satisfactory than earlier Nano Banana models tested in April. However, the generated text misspelled "Forest Festival" twice.
中文摘要
谷歌发布了Nano Banana 2 Lite，正式名称为Gemini 3.1 Flash Lite Image（API标识符为gemini-3.1-flash-lite-image），并宣传其为Gemini图像生成系列中最快且最便宜的模型，专为高速和大规模应用设计。Simon Willison通过Google AI Studio快速测试，用“Where's Waldo”风格提示生成了一幅包含持火腿无线电的浣熊的图像，结果显示比四月测试的早期Nano Banana模型更好，但图中文字两次拼错了“Forest Festival”。
关键要点
Nano Banana 2 Lite is the codename for Gemini 3.1 Flash Lite Image, the fastest and cheapest model in the Gemini image generation lineup.
Nano Banana 2 Lite是Gemini 3.1 Flash Lite Image的代号，是Gemini图像生成系列中最快且最便宜的模型。
The model is accessible via the API as gemini-3.1-flash-lite-image and through Google AI Studio.
该模型可通过API（名称为gemini-3.1-flash-lite-image）和Google AI Studio使用。
In a hands-on test with a complex text-to-image prompt, the output was artistically better than previous Nano Banana models, but it still made spelling mistakes.
在一次使用复杂提示词的实测中，生成的图像在艺术效果上优于之前的Nano Banana模型，但仍存在拼写错误。