Nano Banana API

Google's Gemini 2.5 Flash Image model is built for fast image generation and editing. Use it to blend multiple references, keep character consistency, and apply natural-language transformations for product shots, social creatives, and brand storytelling.

Prompt

A cute cat wearing sunglasses

Parameters

size

auto

Estimated Cost

1.6 Credits

Sample Result

No sample available

Prompt

0 (suggested: 2,000)

Input Images

Upload up to 5 images (max 10MB each)

Click to upload or drag and drop

Supported formats: JPEG, JPG, PNG, WEBP
Maximum file size: 10MB; Maximum files: 5

Size

Click Generate to see preview

History

Max 20 items

0 running · 0 completed

Your generation history will appear here

Nano Banana: fast image generation and editing

Combine text and image inputs to guide composition, then refine with targeted edits and variations. The model keeps latency low so teams can iterate quickly.

Marketing showcase of image generation feature 1

Pricing

Model	Mode	Price
Nano Banana	Image Generation (Gemini 2.5 Flash)	Popular $0.0222/ image(1.6 Credits)

Popular

Nano Banana

Image Generation (Gemini 2.5 Flash)

Price:

$0.0222/ image

(1.6 Credits)

If it's down, we automatically use the next cheapest available—ensuring 99.9% uptime at the best possible price.

What can Nano Banana create?

Product and brand visuals

Nano Banana combines a short brief with one to three reference images to produce clean packshots, hero banners, and lifestyle scenes. It blends style, lighting, and layout so a product or mascot stays consistent across variants, which is useful for catalog refreshes, launch pages, and ad A/B tests. Keep references focused and specify the main subject to improve fidelity and reduce rework.

Try now

Commerce showcase of image generation feature 2

Readable text-on-image

Create posters, menus, labels, and infographics where text must stay readable. A practical workflow is to draft the exact copy first, then render the image with that text so the layout remains stable. If you need images rather than text-only output, say so clearly in the prompt. This flow helps multilingual campaigns ship consistent assets with fewer manual overlays.

Try now

Typography showcase of image generation feature 3

Guided edits and variations

Use Nano Banana for targeted transformations: swap backgrounds, adjust lighting, or replace objects without restarting the concept. It is optimized for fast edits and iterative refinement, so teams can create multiple options quickly while keeping the same character or product identity. This speeds up approval cycles for social creatives, storyboards, and marketing variants that need quick turnarounds.

Try now

Editing showcase of image generation feature 4

Why teams pick Nano Banana

This model combines speed with practical control, so teams can move from brief to usable creative in minutes instead of days.

Clearer text rendering

Improves legibility for labels and headlines inside the image.

Lower-latency workflow

Fast generation makes A/B testing and reviews far smoother.

Provenance signals

Images include SynthID watermarking to support transparency.

How to use Nano Banana

Three simple steps to generate, edit, and deliver images with EvoLink.

Step 1 - Add prompt & references

Write a clear prompt, add style notes, and include one or more reference images to guide composition.

Step 2 - Edit with natural language

Request targeted changes in plain language, then iterate on color, layout, and text until it matches the brief.

Step 3 - Export & reuse

Pick the best variation, export it, and reuse the same character or product settings for consistent campaigns.

Try in playground

Key capabilities

Built for fast, controllable creative work.

Input

Multimodal understanding

The model accepts both text and images, so you can describe the goal while showing references. This makes it easier to match brand style, product details, or scene composition without overlong prompts.

Workflow

Conversational refinement

You can keep the conversation going with short follow-up instructions. Ask for a color shift, a different camera angle, or a new background and the model will update the image instead of starting from scratch.

Control

Character consistency

When you need a series of images with the same character or object, it helps preserve identity across variations. This supports storyboards, campaign sequences, and multi-step tutorials where continuity matters.

Editing

Targeted transformations

Use natural-language edits to replace objects, extend a canvas, or adjust lighting. The model is designed for editing as well as generation, so you can make precise changes with fewer re-renders.

Quality

Real-world knowledge cues

It benefits from Gemini's reasoning and real-world knowledge, which helps it follow common visual logic and avoid awkward details in everyday scenes.

Trust

SynthID watermarking

Images generated in Gemini include an invisible SynthID watermark, providing a provenance signal that the content is AI generated. This supports responsible sharing and platform review workflows.

Frequently Asked Questions

Everything you need to know about the product and billing.

It is the public nickname for Gemini 2.5 Flash Image, Google's image generation and editing model built for low latency. The model creates images from text or combined text-and-image inputs and supports iterative edits. Teams use it for product imagery, social creatives, and storyboards because it balances quality with speed. In EvoLink, the model can be routed through a unified API so you can test it while keeping the same workflow across providers.

The nickname is used by Google for Gemini 2.5 Flash Image in product docs and launch materials. The official model name is still Gemini 2.5 Flash Image, and the nickname refers to the same model rather than a separate product. You might also see "Nano Banana Pro" in documentation for Gemini 3 Pro Image preview. EvoLink keeps the nickname in the UI for clarity while using official model names for routing.

Yes. The model can combine multiple input images into a single output, letting you transfer style, layout, or subject details into one coherent scene. For example, you can upload a product photo and a brand moodboard, then ask it to produce a hero image that matches both. This is especially helpful for maintaining brand consistency across campaigns. For best results, keep inputs focused and limit the number of reference images.

The model supports image editing through natural-language transformations. You can request changes like a new background, a lighting tweak, or a different object without rewriting the entire prompt. Because it is designed for both generation and editing, the iteration cycle is short, which helps teams work faster. Start with a clean base image, then apply small, specific edits step by step so the model can preserve the core subject and style.

Keep the text short, avoid long paragraphs, and specify the hierarchy (headline, subhead, small print). The model performs best when the layout is simple, such as posters, menus, and labels. Provide the exact wording, and ask for clean spacing between lines if you need multiple languages. You can also request a plain background to boost legibility. Iterate with small tweaks rather than changing the entire scene.

Images generated in Gemini include an invisible SynthID watermark, which provides a provenance signal that the content is AI generated. This helps support responsible sharing and reviewer trust. In production workflows, you should still document prompts, references, and approvals internally. The model fits well into compliance processes by combining the SynthID signal with your own asset tracking and review steps.

Google makes Gemini 2.5 Flash Image available to developers through the Gemini API, Google AI Studio, and Vertex AI. EvoLink provides a unified API layer so you can route the model alongside other models without changing your integration. Availability, quotas, and regions depend on your provider account, so check the relevant console for current access. For enterprise governance, Vertex AI is typically the option to review.

Yes. The model is optimized for speed, which makes it practical for teams that need many iterations and approvals. Designers can test multiple concepts quickly, while marketing teams can run A/B creative variations. Store reference images in a shared library and reuse consistent instructions so the brand look stays stable. When feedback arrives, use the edit flow to make incremental changes instead of generating from scratch.

Looking for Nano Banana 2? See the Gemini 3.1 Flash Image model →

API Reference

Select endpoint

Authentication

All APIs require Bearer Token authentication.

Header

Authorization: 
Bearer YOUR_API_KEY

Get API Key

POST

/v1/images/generations

Generate Image

Create an image generation task using text prompts or reference images. Supports text-to-image, image-to-image, and image editing modes.

Asynchronous processing mode, use the returned task ID to .

Generated image links are valid for 24 hours, please save them promptly.

Request Parameters

modelstringRequiredDefault: gemini-2.5-flash-image

Image generation model name.

Examplegemini-2.5-flash-image

promptstringRequired

Prompt describing the image to be generated or how to edit the input image.

Notes

Limited to 2000 tokens

ExampleA cat playing in the grass

sizestringOptionalDefault: auto

Aspect ratio of the generated image.

Value	Description
auto	Auto detect
1:1	Square (1:1)
2:3	Portrait (2:3)
3:2	Landscape (3:2)
4:3	Landscape (4:3)
3:4	Portrait (3:4)
16:9	Horizontal (16:9)
9:16	Vertical (9:16)

Example16:9

image_urlsarrayOptional

Reference image URL list for image-to-image and image editing functions.

Notes

Max 5 images per request
Max size: 10MB per image
Formats: .jpeg, .jpg, .png, .webp
URLs must be directly accessible by the server

Examplehttps://example.com/image1.png

callback_urlstringOptional

HTTPS callback address after task completion.

Notes

Triggered on completion, failure, or cancellation
Sent after billing confirmation
HTTPS only, no internal IPs
Max length: 2048 chars
Timeout: 10s, Max 3 retries

Examplehttps://your-domain.com/webhooks/image-task-completed

Request Example

{
  "model": "gemini-2.5-flash-image",
  "prompt": "A cat playing on the grass",
  "size": "16:9"
}

Response Example

{
  "created": 1757165031,
  "id": "task-unified-1757165031-uyujaw3d",
  "model": "gemini-2.5-flash-image",
  "object": "image.generation.task",
  "progress": 0,
  "status": "pending",
  "task_info": {
    "can_cancel": true,
    "estimated_time": 45
  },
  "type": "image",
  "usage": {
    "billing_rule": "per_call",
    "credits_reserved": 1.6,
    "user_group": "default"
  }
}