Nano Banana API

Nano Banana (Gemini 2.5 Flash Image, model code gemini-2.5-flash-image) is Google's AI image generation and editing model, launched on August 26, 2025, after topping the LMArena image-editing leaderboard as the anonymous codename "nano-banana". It supports natural-language prompts and optional reference images, covering multi-image fusion, character consistency, and precise natural-language retouching, and draws on Gemini's world knowledge to understand complex editing instructions. Every generated or edited image carries an invisible SynthID digital watermark.

The model is offered through the iCreat platform as an API using a two-step asynchronous task flow: submit a task to obtain a task_id, then poll the result endpoint until a terminal status. Output resolution is fixed at 1K, billed per generated image at $0.038 per image; ten output aspect ratios from 1:1 to 21:9 are supported.

Model Positioning

Nano Banana targets general-purpose image production where generation and editing live in one flow. As the first-generation foundation of Google's Nano Banana family, it maintains a character's or object's appearance across generations and edits without fine-tuning, performs targeted local edits from plain natural language, and natively inherits Gemini's world knowledge for instructions that require common sense. Within the family's division of labor, it stands for balanced per-image cost and maturity: Nano Banana Pro pursues flagship fidelity and stronger text rendering, Nano Banana 2 / 2 Lite pursue newer architectures and lower cost, while the original Nano Banana remains a proven, cost-effective choice. Typical needs include character-consistent multi-image creation, reference-based fusion, conversational retouching, and batch brand-asset derivation.

Core Capabilities

Multi-Image Fusion

Blend multiple input images into one: place a product into a new scene, restyle a room by color scheme or texture, or stitch a character into a different shot — all in a single prompt.

Character Consistency

Keep a character's or object's appearance consistent across generations and edits without fine-tuning — the same character in different environments, one product from multiple angles, brand assets unified across images.

Natural-Language Retouching

Supports targeted transformations and local edits: blur the background, remove a stain from a shirt, take a person out of a photo, alter a pose, or colorize a black-and-white photo — each with one sentence of natural language.

Gemini World Knowledge

The model reading your prompt shares Gemini's world knowledge: it can read hand-drawn diagrams, answer real-world questions about images, and execute compound editing instructions in one step — not just patch pixels.

Pricing

Billed per generated image only; the costUSD field in the response returns the actual cost once the task succeeds.

Application Scenarios

• Character-consistent multi-image creation: picture books, comic panels, and IP characters appearing consistently across scenes
• E-commerce and marketing assets: multi-angle product shots, products placed into new backgrounds, batch marketing derivatives
• Conversational retouching: object removal, background swaps, pose changes, and colorization via natural language
• Templated design assets: real-estate cards, employee badges, and product catalogs generated from one template
• Educational illustrations: diagrams and explainer graphics generated or edited with world knowledge

Model Comparison

Same-Series Comparison

Cross-Model Comparison

Same-series data comes from Google's official documentation and release pages; this model's pricing follows the iCreat channel price table.

Why Choose Nano Banana?

• Generation and editing share one pipeline — no switching between image-creation and retouching endpoints
• Character consistency without fine-tuning, keeping characters and products unified across images
• Local, precise edits from plain natural language with a minimal learning curve
• Gemini's world knowledge built in, for editing instructions that require common sense
• A proven, massively validated foundation of the family, balancing per-image cost and quality

Specifications

Architecture

Nano Banana (Gemini 2.5 Flash Image) belongs to the native image generation branch of Google's Gemini 2.5 family. It is not a diffusion model bolted onto Gemini — it is part of Gemini's multimodal architecture: the same model that reads the prompt holds the world knowledge, and text and images interleave within one context, so generation and editing share a single pipeline. Output images are billed at a fixed 1,290 output tokens ($30 per million tokens on the official API, roughly $0.039 per image). Every generated or edited image carries an invisible SynthID digital watermark identifying it as AI-generated.

Notes

• Submit and query must be chained with the same task_id
• While processing, result is []; keep polling, with a suggested interval of 2–5 seconds
• FAILED is terminal; check the prompt, reference image URLs, image format, and aspect ratio before resubmitting
• image is optional: prompt-only requests generate new images; with reference images, the model edits or generates from references
• image_size currently supports 1K only
• aspect_ratio is required and must be one of the ten allowed values
• Every item in the image array must be a publicly accessible image URL
• Output images carry an invisible SynthID watermark (always on for Google; cannot be disabled)
• costUSD is returned only when the task succeeds; failed tasks do not produce a cost field

FAQ

How do I get started with the Nano Banana API?​

Register on the iCreat platform and obtain an API Key from the console, send a generation request to the submit endpoint, then poll the result endpoint with the returned task_id. All requests are authenticated with the Authorization: Bearer header. Keep your API Key safe and never expose it in client code or public repositories.

How is the cost calculated?​

Billed per generated image: $0.038 per image at 1K resolution. For example, generating 10 images costs $0.38. The costUSD field in the response gives the actual cost once the task succeeds.

Does it support generation or editing?​

Both, on the same submit endpoint: with only a prompt, it generates new images; with the image reference array, it generates from references or applies local edits to the images (object removal, background swaps, pose changes, and more). For multi-turn editing, submit the previous result as the next reference image.

What resolutions and aspect ratios are supported?​

Resolution currently supports 1K only. Aspect ratios support ten values: 1:1, 2:3, 3:2, 3:4, 4:3, 4:5, 5:4, 9:16, 16:9, and 21:9 — the ratio must be specified explicitly. Pick the right size for each delivery platform.

How do I query the task result?​

Send a POST request to https://api.icreat.ai/v1/task/result with the task_id in the body. A status of SUBMITTED means processing, with result as [] — keep polling. When status is SUCCEEDED, result returns the image resource array; read url or download_url to retrieve the image.

What if the task fails?​

FAILED is a terminal status. Check in order: whether the prompt is empty, whether reference image URLs are publicly accessible, whether image_size is 1K, and whether aspect_ratio is one of the ten allowed values. Fix any issue and resubmit. The cost field is returned only when the task succeeds.