Nano Banana 2 · Text to Image
fal-ai/nano-banana-2Generate images from text with Nano Banana 2 · Text to Image from fal through one Spiral API key. Generates up to 4 images from a text prompt with Google's Nano Banana 2, at 0.5K to 4K, with optional web search grounding. It runs in the background and returns images. Pay as you go, with no subscription.
Pricing
$0.104 per image at 1K ($0.078 at 0.5K, $0.156 at 2K, $0.208 at 4K). Web search: $0.0195 per request. High thinking: $0.0026 per request.
Limits
- Runs
- In the background
- Returns
- image
- Longest queue wait
- 1 hour
- Run deadline
- 30 minutes
- Longest run, end to end
- 225 minutes
- Largest inline input
- 1 MB
- Largest input file
- 2 GB
- Largest output
- 5 GB
Input
| Field | Kind | Default | Description |
|---|---|---|---|
prompt required | text | — | The text prompt to generate an image from. |
history | JSON | — | Previous native Gemini turns, including original parts and thought signatures from conversations. The current prompt starts a new user turn. |
context | JSON | — | Ordered text, image, PDF, UTF-8 text-file or video parts appended to the current prompt. Original media bytes are preserved; images count toward the total 14-reference limit. |
return_history | true or false | false | Return reusable native conversations with original image parts and provider thought signatures. Uses a native provider. Each generated conversation is independent; send its messages as history for the next turn. |
thinking_level | one of minimal, high | — | Controls thinking effort ('minimal' or 'high') and requests thought summaries. Omit to use the provider default ('minimal'); thinking cannot be disabled. |
image_passthrough | true or false | false | Send reference images and return provider images without resizing or re-encoding, preserving transparency and metadata. Uses direct Google or Vertex APIs; the returned image format may differ from output_format. |
include_thoughts | true or false | — | Whether to include thought summaries. Omit to preserve the existing behavior: an explicit thinking_level requests summaries. Set limit_generations=false to return summary text in description. This does not disable thinking or return intermediate thought images. |
temperature | number 0 to 2 | — | Sampling temperature. Omit to use the provider default of 1. Explicit values use a direct Google or Vertex API. |
top_p | number 0 to 1 | — | Nucleus sampling probability. Omit to use the provider default of 0.95. Explicit values use a direct Google or Vertex API. |
max_output_tokens | number (whole) 1 to 32768 | — | Maximum output tokens for the generation. Omit to retain the provider default. Explicit values use a direct Google or Vertex API. |
num_images | number (whole) 1 to 4 | 1 | The number of images to generate. |
seed | number (whole) | — | The seed for the random number generator. |
aspect_ratio | one of 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 | auto | The aspect ratio of the generated image. Supports extreme ratios: 4:1, 1:4, 8:1, 1:8. Use "auto" to let the model decide based on the prompt. |
output_format | one of jpeg, png, webp | png | The format of the generated image. |
safety_tolerance | one of 1, 2, 3, 4, 5, 6 | 4 | The safety tolerance level for content moderation. 1 is the most strict (blocks most content), 6 is the least strict. |
system_prompt | text | 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 | one of 0.5K, 1K, 2K, 4K | 1K | The resolution of the image to generate. |
limit_generations | true or false | true | Experimental parameter to limit the number of generations from each round of prompting to 1. Set to |
enable_web_search | true or false | false | Enable web search for the image generation task. This will allow the model to use the latest information from the web to generate the image. |
enable_image_search | true or false | false | Ground image generation with Google Image Search. Can be combined with enable_web_search; source attribution is returned in grounding_metadata. Uses direct Google or Vertex APIs. |
Same request in code
curl https://api.spiral.ai/v1/runs \
-H "Authorization: Bearer $SPIRAL_API_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: $(uuidgen)" \
-d '{
"provider": "fal",
"model": "fal-ai/nano-banana-2",
"input": {
"prompt": "An action shot of a black lab swimming in an inground suburban swimming pool. The camera is placed meticulously on the water line, dividing the image in half,…"
}
}'Frequently asked questions
How much does Nano Banana 2 · Text to Image cost?
$0.104 per image at 1K ($0.078 at 0.5K, $0.156 at 2K, $0.208 at 4K). Web search: $0.0195 per request. High thinking: $0.0026 per request. You pay as you go from one Spiral balance, with no subscription, and a failed run is not charged.
How do I use Nano Banana 2 · Text to Image through an API?
Send `POST /v1/runs` with `"provider": "fal"`, `"model": "fal-ai/nano-banana-2"` and the input fields on this page, using one Spiral API key. It runs in the background: poll the run, or name a webhook to be told when it finishes.
Do I need an account with fal to use Nano Banana 2 · Text to Image?
No. Spiral runs it on its own fal account and charges your Spiral balance.
Can my AI agent use Nano Banana 2 · Text to Image?
Yes. Agents find and run it through Spiral's MCP server, CLI or agent skill, with the same API key.