Generate images with Nano Banana 2 — fast, high-fidelity output combining speed with 4K resolution and strong character consistency. Describe any scene and get a polished image in seconds. Try free — pay per generation, no subscription.
About this model
Nano Banana 2, technically known as Gemini 3.1 Flash Image, represents the pinnacle of Google's AI image generation technology. As a flagship text-to-image model, it merges the depth of Pro models with the lightning speed of Flash variants. This model introduces Revolutionary Character Consistency, allowing creators to maintain perfect facial features, clothing, and styling for consistent storytelling. With support for high-fidelity 4K resolution, precise advanced text rendering, and deep Google AI integration, Nano Banana 2 is optimized for professional commercial workflows, digital art, and immersive visual branding.
Cost analysis
| Provider | Cost | Notes |
|---|---|---|
| muapiapp | $0.06 (1k) | $0.09 (2k) | $0.12 (4k) | muapiapp provides the most competitive pricing for high-resolution Google AI generation, saving users up to 25% compared to other providers. |
| Competitors | $0.08 (1k) | $0.12 (2k) | $0.16 (4k) | Standard industry rates are typically 25% higher for the same level of resolution and quality provided by the Gemini 3.1 Flash architecture. |
muapiapp provides the most competitive pricing for high-resolution Google AI generation, saving users up to 25% compared to other providers.
Standard industry rates are typically 25% higher for the same level of resolution and quality provided by the Gemini 3.1 Flash architecture.
* Competitor pricing is estimated based on similar model architectures and usage tiers.
Configuration schema
| Parameter | Type | Description | Default |
|---|---|---|---|
| Prompt | string | Text prompt describing the desired image content. | A playful, vibrant fruit market scene with bright natural lighting and rich colors. In the center, giant bananas are arranged carefully to form bold, physical letters, stacked and curved naturally, with realistic texture, yellow peel details, and soft shadows on the ground. The bananas look fresh, glossy, and dimensional, like real objects sculpted into typography. The camera angle is slightly low and centered to make the text feel large and dominant.
The banana-formed letters clearly spell:
Nano Banana 2
Ultra-realistic detail, cinematic lighting, sharp focus, fun and memorable composition, highly readable fruit typography, premium poster-style presentation where the name itself is the main visual element. |
| Aspect Ratio | Enum (15 options) | The aspect ratio of the generated image. | 1:1 |
| Google Search | boolean | Whether to use Google Search for prompt enhancement. | false |
| Resolution | Enum (3 options) | The resolution of the generated image. | 1k |
| Output Format | Enum (2 options) | The format of the output image. | jpg |
Text prompt describing the desired image content.
A playful, vibrant fruit market scene with bright natural lighting and rich colors. In the center, giant bananas are arranged carefully to form bold, physical letters, stacked and curved naturally, with realistic texture, yellow peel details, and soft shadows on the ground. The bananas look fresh, glossy, and dimensional, like real objects sculpted into typography. The camera angle is slightly low and centered to make the text feel large and dominant.
The banana-formed letters clearly spell:
Nano Banana 2
Ultra-realistic detail, cinematic lighting, sharp focus, fun and memorable composition, highly readable fruit typography, premium poster-style presentation where the name itself is the main visual element.The aspect ratio of the generated image.
1:1Whether to use Google Search for prompt enhancement.
falseThe resolution of the generated image.
1kThe format of the output image.
jpgDeveloper documentation
Frequently asked
It is Google's latest Gemini 3.1 Flash Image model, offering a unique balance of extreme speed, 4K resolution, and industry-leading character consistency.
It uses advanced latent space tracking to preserve facial features, clothing, and unique identifiers across a sequence of generated images.
The model supports high-fideity generation from 512px up to native 4K across all standard aspect ratios.