Nano Banana 2
Nano Banana 2 (Gemini 3.1 Flash Image) is Google’s next-generation AI model for image editing and generation, making visual creation as simple and intuitive as describing it in words. Built on Google’s cutting-edge computer vision and generative AI technologies, it combines precise control, creative flexibility, and deep semantic understanding to deliver professional-grade image editing and generation.
Related Models
Nano Banana 2 Text-to-Image
Nano Banana 2 Text-to-Image (Gemini 3.1 Flash Image) is Google’s next-generation AI model for image editing and generation, making visual creation as simple and intuitive as describing it in words. Built on Google’s cutting-edge computer vision and generative AI technologies, it combines precise control, creative flexibility, and deep semantic understanding to deliver professional-grade image editing and generation.
Read Me
Google Nano Banana 2 API
Introduction
Nano Banana 2 is Google’s balanced image generation and editing model for everyday production, repeated creative work, and cost-conscious visual teams. It combines strong instruction following, readable text, multi-reference input, and outputs up to 4K without the premium positioning of Nano Banana Pro.
On iCreat, you can generate new images from text, upload references for guided creation, or edit existing visuals through the same model page. The endpoint supports up to 11 reference images, including six object images and five portrait photos. Choose from 1K, 2K, or 4K output across ten common aspect ratios, then use the model through the online Playground or the google/gemini-3-1-flash-image API.
Key Features
Balanced image quality and efficiency
Nano Banana 2 balances image quality, instruction following, and production cost for everyday visual work. It suits teams that need dependable generation and editing across repeated campaigns, social assets, product scenes, and content variations without paying the premium required for the most demanding professional design workflows or final presentation assets.
Multi-reference generation and editing
Upload up to 11 reference images, including as many as six object images and five portrait photos. This lets users combine products, packaging, people, clothing, environments, and styles in one request while keeping the reference workflow structured enough for repeated generation, editing, localization, and campaign asset variation.
Readable text and controlled layouts
Nano Banana 2 can place readable text inside posters, menus, product labels, social graphics, and simple infographics. It works best when the prompt specifies exact wording, position, hierarchy, and surrounding elements, helping users create practical visual assets that need both imagery and organized written content for real publishing workflows.
Natural-language image editing
Use prompts and reference images to revise existing visuals through the same model page. Nano Banana 2 can change backgrounds, lighting, props, styling, or scene details while preserving the main subject when clearly instructed, making it useful for campaign variations, localization, seasonal updates, and everyday asset maintenance.
Affordable 1K to 4K production
Nano Banana 2 offers fixed pricing by resolution: $0.07 for 1K, $0.105 for 2K, and $0.14 for 4K. With no separate quality tier to choose, teams can estimate per-image costs quickly and scale routine production across multiple formats, channels, campaign versions, output ratios, and final delivery sizes.
How to Use
1. Choose your workflow
Enter a prompt without uploading an image for text-to-image generation. Upload one or more references when you want to guide a new composition or edit an existing asset.
2. Describe the result
State the subject, setting, layout, text, lighting, and visual style. For edits, clearly separate what should change from the details that must remain consistent.
3. Add reference images
Upload up to 11 references and explain each image’s role, such as product shape, packaging, portrait identity, clothing, environment, or lighting.
4. Select ratio and resolution
Choose a supported aspect ratio and select 1K, 2K, or 4K resolution. Generate in the Playground or call the google/gemini-3-1-flash-image API.
Pricing
Nano Banana 2 uses fixed per-image pricing based on resolution for text-to-image generation, reference-based creation, and image editing.
| Resolution | Unit Price |
|---|---|
| 1K | $0.07/image |
| 2K | $0.105/image |
| 4K | $0.14/image |
A 4K Nano Banana 2 image costs the same as a 1K or 2K Nano Banana Pro image on iCreat.
Model Comparison
Nano Banana 2 vs Nano Banana Pro
Choose Nano Banana 2 for routine production and Nano Banana Pro for the most demanding professional visual briefs.
| What matters | Nano Banana 2 | Nano Banana Pro |
|---|---|---|
| Best for | Daily content production | Professional final assets |
| Complex briefs | Strong | Stronger |
| Brand consistency | Good | Better suited |
| Reference structure | 6 objects + 5 portraits | 6 objects + 5 portraits |
| Reference limit | Up to 11 | Up to 11 |
| Prompt limit | 8,192 characters | 8,192 characters |
| Output control | Ratio + resolution | Ratio + resolution |
| Quality tiers | None | None |
| 1K price | $0.07 | $0.14 |
| 2K price | $0.105 | $0.14 |
| 4K price | $0.14 | $0.24 |
| Budget planning | Lower-cost | Premium |
Nano Banana 2 vs GPT Image 2
Nano Banana 2 offers simpler fixed pricing, while GPT Image 2 provides more reference capacity and finer quality controls.
| What matters | Nano Banana 2 | GPT Image 2 |
|---|---|---|
| Best for | Daily visual production | Text-heavy complex briefs |
| Text and layout | Strong | Strong |
| Editing workflow | Prompt + references | Prompt + references |
| Reference structure | 6 objects + 5 portraits | General references |
| Reference limit | Up to 11 | Up to 16 |
| Prompt limit | 8,192 characters | 32,000 characters |
| Output control | Ratio + resolution | Exact size + quality |
| Quality tiers | None | Low / Medium / High |
| 1K price | $0.07 | $0.03–$0.22 |
| 2K price | $0.105 | $0.06–$0.44 |
| 4K price | $0.14 | $0.09–$0.66 |
| Budget planning | Fixed by resolution | Flexible by quality |
Parameters
| Parameter | Type | Required | Description |
|---|---|---|---|
prompt |
string | Yes | Image instruction with a maximum length of 8,192 characters |
image |
string[] | No | Up to 11 reference-image URLs for guided generation or editing |
aspect_ratio |
string | Yes | Selects one of ten supported output ratios |
image_size |
string | Yes | Selects 1K, 2K, or 4K resolution |
| Specification | Value |
|---|---|
| Model ID | google/gemini-3-1-flash-image |
| Workflows | Text-to-image, reference generation, image editing |
| Reference images | Up to 11 |
| Object references | Up to 6 |
| Portrait references | Up to 5 |
| Input formats | PNG, JPG, JPEG, WebP |
| Resolution tiers | 1K, 2K, 4K |
| Aspect ratios | 1:1, 2:3, 3:2, 3:4, 4:3, 4:5, 5:4, 9:16, 16:9, 21:9 |
| Billing unit | USD per image |
| API pattern | Asynchronous task submission |
Use Cases
High-volume campaign variations
Create multiple versions of one campaign concept across formats, backgrounds, seasons, messages, and audience segments while keeping the main product, person, or visual direction consistent.
Social media and advertising assets
Generate square, vertical, horizontal, and ultrawide visuals for social posts, display ads, stories, and promotional campaigns using one prompt and a consistent set of references.
Product and lifestyle visuals
Combine product, packaging, environment, lighting, and style references to place an existing item into new commercial, editorial, or lifestyle scenes without rebuilding the composition manually.
Character-based content
Use portrait, clothing, pose, background, and style references to create recurring people across social content, campaign concepts, editorial scenes, lifestyle imagery, and localized visual variations.
Menus, posters, and infographics
Create visuals that combine readable text, hierarchy, icons, labels, and supporting imagery for menus, posters, simple infographics, announcements, educational content, and other information-led design assets.
Image editing and localization
Change backgrounds, props, lighting, language, seasonal details, or regional cues while preserving the original subject, making existing assets easier to adapt for different markets and campaigns.
Prompt Tips
Assign every reference a role. State which image controls the product, person, packaging, clothing, background, lighting, or style.
Match the ratio to the channel. Choose the final format before writing the composition instructions so the subject, text, and negative space fit the intended placement.
Quote required text. Put exact wording in quotation marks and specify its position, hierarchy, and relationship to surrounding elements.
Separate changes from preserved details. Identify what should change and what must remain consistent, especially when editing products, people, packaging, or branded assets.
Notes
- The current iCreat endpoint supports 1K, 2K, and 4K output, but it does not expose Google’s native 0.5K option.
- The 11-image limit includes up to six object references and five portrait photos; it does not allow 11 references of either type without restriction.
- Google Search grounding, Thinking controls, and video-to-image input are not exposed through the current iCreat parameters.
- Images generated by the Nano Banana model family include SynthID watermarking.
Related Models
Choose it for complex professional assets, stronger creative control, and brand-focused production where precision matters more than per-image cost.
Choose it for longer prompts, up to 16 general references, exact output dimensions, and flexible Low, Medium, and High quality pricing.