# Nano Banana 2 (Edit) - Core.Today AI API

> Edit endpoint for Nano Banana 2, built on Gemini 3.1 Flash Image. Adds resolution control (1K/2K/4K), Google Search grounding, and thinking mode while preserving conversational editing and multi-image fusion.

- **Provider**: Google
- **Model ID**: fal-ai/nano-banana-2/edit
- **Category**: Image Generation
- **Credits**: 190 per image (1K); 280 for 2K, 370 for 4K
- **Speed**: Medium
- **Quality**: Ultra

## Features
- Resolution control: 1K, 2K, 4K
- Optional web search grounding
- Thinking mode (minimal/high)
- Multi-image fusion and conversational editing
- Character consistency across edits

## Use Cases
- Higher-resolution edited renders
- Information-grounded illustration (web search)
- Multi-reference scene composition
- Multilingual typography edits

## API Endpoint

Base URL: https://api.core.today/v1
Create Prediction: POST /predictions
Get Status: GET /predictions/{job_id}

## Authentication
Header: X-API-Key: YOUR_API_KEY

## Input Parameters

### Required
- **prompt**: string - How you want to edit the image

### Optional
- **num_images**: integer (default: 1) - Number of images to generate
- **seed**: integer - The seed for the random number generator.
- **aspect_ratio**: string (default: auto) - Output aspect ratio (auto lets the model decide)
  Options: auto, 21:9, 16:9, 3:2, 4:3, 5:4, 1:1, 4:5, 3:4, 2:3, 9:16, 4:1, 1:4, 8:1, 1:8
- **output_format**: string (default: png) - Output image format
  Options: jpeg, png, webp
- **safety_tolerance**: string (default: 4) - The safety tolerance level for content moderation. 1 is the most strict (blocks most content), 6 is the least strict.
  Options: 1, 2, 3, 4, 5, 6
- **sync_mode**: boolean (default: false) - If `True`, the media will be returned as a data URI and the output data won't be available in the request history.
- **image_urls**: array - Input image URLs to edit
- **system_prompt**: string - Optional system instruction that steers the model's persona and output style across the request. Leave blank to omit; when provided, it is sent as the system instruction to Gemini (or as a system message on OpenAI-compatible providers).
- **resolution**: string (default: 1K) - Output resolution — pricing scales with resolution
  Options: 0.5K, 1K, 2K, 4K
- **video_url**: string - Optional video to use as input context. Accepts an http(s)/data URL to a video file (downloaded and sent inline, max 15MB) or a YouTube URL (passed through to the model without downloading).
- **audio_url**: string - Optional audio file to use as input context. Accepts an http(s)/data URL (downloaded and sent inline, max 15MB).
- **pdf_url**: string - Optional PDF document to use as input context. Accepts an http(s)/data URL (downloaded and sent inline, max 15MB).
- **limit_generations**: boolean (default: true) - Experimental parameter to limit the number of generations from each round of prompting to 1. Set to `True` to to disregard any instructions in the prompt regarding the number of images to generate and ignore any intermediate images generated by the model. This may affect generation quality.
- **enable_web_search**: boolean (default: false) - Enable web search grounding for real-time information
- **thinking_level**: string - Enable model thinking (omit to disable)
  Options: minimal, high

## Examples

### Multi-image Composition
Compose a scene from multiple input photos
```json
{
  "model": "fal-ai/nano-banana-2/edit",
  "input": {
    "prompt": "make a photo of the man driving the car down the california coastline",
    "image_urls": [
      "https://storage.googleapis.com/falserverless/example_inputs/nano-banana-edit-input.png",
      "https://storage.googleapis.com/falserverless/example_inputs/nano-banana-edit-input-2.png"
    ],
    "aspect_ratio": "auto",
    "resolution": "1K",
    "output_format": "png",
    "num_images": 1
  }
}
```

## Response Format
```json
{
  "job_id": "abc123",
  "status": "pending | processing | completed | failed",
  "result": "URL or data (when completed)"
}
```

## Usage Flow
1. POST /predictions with model and input -> receive job_id
2. GET /predictions/{job_id} -> poll until status is completed or failed
3. Result contains output URL(s)

## Tips
- Resolution maps directly to credit cost — start at 1K for iteration
- Pass multiple image_urls to compose subjects from different sources
- Enable thinking_level=high for hard editing instructions
- Use enable_web_search when you need up-to-date facts in the image