falImageAvailable now

google/nano-banana-2.1

google/nano-banana-2.1 is an image tool from fal that runs in the background and returns images. Pay as you go, with one API key.

Pricing

$0.0514 per 1K image ($0.0771 at 2K, $0.173 at 4K). Prompt: $2.29 per 1M tokens. Reasoning: $11.47 per 1M tokens.

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
500 MB
Largest output
5 GB
Outputs per run
16

Input

FieldKindDefaultDescription
prompt requiredtext—

The text prompt to generate an image from.

num_imagesnumber (whole)
1 to 4
1

The number of images to generate.

seednumber (whole)—

The seed for the random number generator.

aspect_ratioone 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_formatone of
jpeg, png, webp
png

The format of the generated image.

safety_toleranceone 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_prompttext

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).

resolutionone of
1K, 2K, 4K
1K

The resolution of the image to generate.

limit_generationstrue or falsetrue

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_searchtrue or falsefalse

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.

thinking_levelone of
minimal, medium, high
medium

How much the model reasons before generating. Higher levels can improve complex compositions and use more thinking tokens.

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": "google/nano-banana-2.1",
    "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,…"
    }
  }'