
DIY infographic
16:9High-quality flat lay photography creating a DIY infographic that simply explains how the water cycle works, with hand-drawn arrows and labels.
google/nano-banana-2Nano Banana 2 image generation API by Google — Gemini 3.1 Flash with 4K output, 15 aspect ratios, and character consistency for up to 5 characters.
curl https://api.runbase.net/v1/runs \
-H "Authorization: Bearer $RUNBASE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "google/nano-banana-2",
"input": {
"prompt": "A cinematic product photo of a ceramic lamp",
"aspect_ratio": "1:1",
"resolution": "1K"
}
}'
High-quality flat lay photography creating a DIY infographic that simply explains how the water cycle works, with hand-drawn arrows and labels.

A three-panel comic-style comparison of Cumulus, Stratus, and Cirrus clouds in dramatic skies, each panel with a bold legible label.

A four-panel collage of the same kangaroo character in a consistent outfit across vacation scenes — boating, a café, a hotel room, and a bouncy castle.
Nano Banana 2 is the most versatile model in Google's Nano Banana family, released in February 2026. Built on Gemini 3.1 Flash, it delivers Pro-level quality at Flash speed. It supports 15 aspect ratios — the widest selection available — including ultra-wide options (8:1, 1:8) and auto. Output resolution goes up to 4K. It accepts up to 14 reference images at 30 MB each and can maintain character consistency for up to 5 distinct characters across images.
High-volume brand asset production at scale. Character-consistent workflows — comics, storyboards, mascot campaigns. Infographics and diagrams with accurate text rendering. Ultra-wide banners and panoramic compositions using 8:1 or 21:9 ratios. Reference-guided editing with up to 14 images.
All parameters are passed in the input object of the run request.
| Parameter | Required | Description |
|---|---|---|
| prompt | Yes | Text description (1–20000 chars) |
| aspect_ratio | No | Output aspect ratio. Default 1:1. Supports 15 ratios including ultra-wide (8:1, 1:8) and auto |
| resolution | No | Output resolution. Default 1K. Options: 1K, 2K, 4K |
| image_urls | No | Up to 14 reference images (max 30 MB each) for guided generation |
When using character consistency, give each character a distinct visual description and refer to them consistently across prompts. "The red-haired fox in a blue scarf" should stay the same phrase each time.
1K is fast and cheap for iteration. 2K balances quality and cost. Reserve 4K for final deliverables — the price difference is significant.
8:1 and 1:8 are powerful for banners and vertical scrolls, but they require prompts that fill the space. Describe a scene with enough elements to avoid sparse compositions.
Fifteen options: 1:1, 1:4, 1:8, 2:3, 3:2, 3:4, 4:1, 4:3, 4:5, 5:4, 8:1, 9:16, 16:9, 21:9, and auto. The default is 1:1.
1K (default), 2K, and 4K. Each resolution tier has a different price point.
Describe up to 5 characters with distinct visual traits and the model will maintain their appearance across generated images. Use reference images to further anchor each character's look.