Nano banana

Google launched a series of graphics models

Role Consistency

Multigraph Fusion

Text Rendering

Old Photo Restoration & Enhancement

Multimodal Model

Model Evaluation & Comparison

Inference Optimization

Nano Banana (Gemini 2.5 Flash Image) is Google's most efficient model for fast generation and character consistency, while Nano Banana Pro (Gemini 3 Pro Image) is a professional flagship that uses Thinking to achieve perfect text rendering and accurate compliance with complex instructions.

Preview

Introduction

functional characteristics

Nano Banana (Gemini 2.5 Flash Image) supports text and image mixed input, enabling character consistency generation, multi-image fusion and natural language-driven precision editing.
Nano Banana Pro (Gemini 3 Pro Image) introduces thought-chain reasoning, generates 2K/4K high-resolution images, has perfect text rendering, multilingual support and Google-based fact calibration capabilities.

application scenarios

E-commerce product diagram generation: keep the appearance of goods consistent, and quickly make multi-scene display diagrams.
Old photo restoration and style transfer: color old black and white photos, or convert them into portraits of different ages and artistic styles.
Infographics and marketing material creation: Combine real-time data to create weather cards, recipe illustrations or social media posters with accurate text.
Comics and game asset creation: turn characters into hand models, continue story storyboards, or generate multi-perspective consistent game originals.

suitable for the croud

Content creators, designers, e-commerce operators and other professional users who need to produce visual content efficiently.
Students, teachers and general users, for learning records, creative expression or daily revision.

type of pricing

Free users have a fixed amount of generation per day (such as 100 times in Gemini App), and after excess, they need to subscribe to a paid plan;API charges by output token, Nano Banana about $0.039/piece, Nano Banana Pro 4K about $0.24/piece.

supported langauges

Support multilingual text rendering, including Chinese, English, Korean, etc., which can be directly translated in the image and retain the original design style.

technical highlights

The bottom layer is Google's self-developed Gemini series multi-modal large model, Nano Banana is based on Gemini 2.5 Flash, Pro version is based on Gemini 3 Pro, introducing reasoning mechanism to improve factual accuracy.
In the LMArena image editing list, it surpasses competitors such as GPT-4o Image, especially in character consistency and text rendering ability.

usage mode

It can be used directly through Gemini official website, Google AI Studio web and mobile apps, and is also integrated into third-party platforms such as Lovart and Baidu Seconds.

update history

The first generation Nano Banana (Gemini 2.5 Flash Image) was released in August 2025, focusing on fast generation and character consistency.
Nano Banana Pro (Gemini 3 Pro Image) launched in November 2025 with 4K output, multilingual text rendering and search enhancements.