구글, 제미나이에 ‘나노바나나2’ 출시

Published: (February 27, 2026 at 12:58 AM EST)
2 min read

Source: Byline Network

Release Announcement

Google announced on the 27th that it will launch Gemini 3.1 Flash, featuring the latest image‑generation model NanoBanana 2 (Gemini 3.1 Flash image). The NanoBanana series had previously been based on the Gemini 3.0 model.

Image Example

NanoBanana 2 generated image – Prompt: “Create an image of 14 characters and objects having fun on a farm. The overall mood should be fun, lively, and joyful. It is crucial to keep the identity of all 14 characters and objects consistent.”

Key Features

  • Performance: Delivers results comparable to NanoBanana Pro while maintaining the speed of the standard version.
  • Expert Functions: Many expert‑only capabilities have been migrated to the general model.
  • High‑Speed Intelligence: Gemini Flash’s rapid intelligence functions are applied to video production, enabling fast editing and repetitive tasks.
  • Real‑World Knowledge: Leverages Gemini LLM’s real‑world knowledge base and incorporates real‑time information and images from web search.
  • Deep Understanding: Can create infographics, convert notes into diagrams, and generate digital visualizations.
  • Text Generation & Localization: Generates precise text for marketing mock‑ups or greeting cards and can translate/localize text within images.
  • Consistency & Fidelity: In a single workflow, maintains similarity for up to 5 characters and factuality for up to 14 objects without altering input shapes, allowing storyboard creation and narrative structuring.
  • Improved Instruction Following: More strictly adheres to complex requests and captures subtle idea variations.
  • Resolution & Aspect Ratios: Supports resolutions from 512 px up to 4K and a variety of aspect ratios, with enhanced visual quality.

Model Integration & Access

  • NanoBanana 2 replaces the previous NanoBanana Pro across Gemini app models such as Fast, Thinking, and Pro.
  • Google AI Pro and Ultra subscribers can continue using NanoBanana Pro for specialized tasks.
  • The model is accessible via Google’s AI Mode, Lens, apps, and browsers.
  • Available for preview in AI Studio and the Gemini API.
  • Also reachable through Google Cloud Vertex AI.
  • Flow users can set NanoBanana 2 as the default image‑generation model and use it without consuming credits.
0 views
Back to Blog

Related posts

Read more »