Nano Banana 2: Google rebuilds its viral image model around consistency and text
In one sentence Google releases Nano Banana 2 (aka Gemini 3.1 Flash Image): much better text rendering, consistency for up to 5 characters and 14 objects, default for image generation in Gemini app, Flow, Lens, and Search AI Mode.
Nano Banana is Google's image-generation model that went viral in 2025 because it was fast, free, and produced surprisingly consistent images. On February 26, 2026 Google releases version 2, which is technically Gemini 3.1 Flash Image.
What it improves over v1:
- text inside images is far more legible (historically the weak point of image-gen models);
- it holds up to 5 distinct characters across multiple scenes without mixing them up (useful for storytelling);
- up to 14 distinct objects handled coherently in one workflow.
It becomes the default for generating images in the Gemini app, in the Flow video tool, in Google Lens, and in Google Search's new "AI Mode" across 141 countries. For content creators, it's the most powerful "free-under-your-fingers" model in the consumer market at release time.
Companies
Google, Google DeepMind
Tools
Nano Banana 2, Gemini 3.1 Flash Image
Tags
Sources