Google Nano Banana Pro Text To Image Multi API Documentation

Google Nano Banana Pro Text To Image Multi API Documentation

Playground

Google’s Nano Banana Pro (Gemini 3.0 Pro Image) is a next-generation text-to-image model capable of generating multiple high-quality images in a single run. Extremely low cost — only $0.07 per image. Ready-to-use REST inference API, best performance, no coldstarts, affordable pricing.

Features

Nano Banana Pro Text-to-Image Multi (Gemini 3.0 Pro Image) is Google’s next-generation text-to-image model with true multi-image generation. One prompt generates two images in a single run, perfect for rapid exploration, A/B testing, and storyboard creation. On WaveSpeedAI, this endpoint costs $0.07 per image, or $0.14 per request for two images.

• Need single-image hero shots? Try Nano Banana Pro Text-to-Image
• Need higher resolution? Try Nano Banana Pro Text-to-Image Ultra
• Need image editing? Try Nano Banana Pro Edit

🌟 Why it stands out

• True multi-image generation Generate two images from one prompt in a single request. num_images is fixed at 2.

• Two-image output Each request returns two images, with controls for aspect ratio and output format.

• Powerful prompt understanding Handles detailed, editorial-style prompts with nuanced control over subject, composition, lighting, and mood.

• Aspect ratio and format flexibility Supports 3:2, 2:3, 3:4, and 4:3 aspect ratios, with png or jpeg output.

• Fast and affordable at scale Batched generation plus a flat per-image price makes it ideal for campaigns, design sprints, and large creative pipelines.

⚙️ Capabilities

• Input: one natural-language text prompt.
• Output: two images per request.
• num_images: fixed at 2; each request generates two images.
• aspect_ratio: 3:2, 2:3, 3:4, or 4:3. Default: 3:2.
• output_format: png or jpeg. Default: png.

Example use cases:

• Generate two fashion shots from one art-direction prompt.
• Produce multiple thumbnail variants for a single YouTube or e-commerce listing.
• Create quick moodboard frames that share the same style but differ in composition.

💰 Pricing

• $0.07 per image × 2 images = $0.14 per request.

💡 Best Use Cases

• A/B testing ad creatives Produce multiple variations of a campaign visual, then pick the best-performing one.

• Thumbnail and banner exploration Generate two layout and color options for social posts, storefronts, and landing pages.

• Storyboards and keyframes Create multiple frames from a shared prompt to sketch out sequences for video, animation, or presentations.

• Brand style development Lock in a consistent stylistic direction while exploring slight changes in pose, background, or framing.

• Concept art and moodboards Batch-generate different interpretations of a scene to speed up early-stage visual ideation.

📝 Notes

• Prompts must comply with Google’s safety policies. If a request fails or content is removed, review the prompt for restricted material and rephrase accordingly.
• Each request generates two images. num_images cannot be changed to another value.
• Avoid conflicting instructions inside a single prompt (for example, “night” and “bright midday sun” together), as this can reduce quality or coherence.
• Higher-level, descriptive prompts usually work better than extremely short or keyword-only inputs.

🌏 Where Nano Banana Pro Fits In

Use Google Nano Banana Pro Text-to-Image Multi when:

• You want two image options from a single prompt in one run.
• You are iterating on creative direction and need fast, low-cost visual diversity.
• You are building tools that benefit from batched image generation for performance and pricing reasons.

For complementary workflows on WaveSpeedAI:

• Nano Banana Pro Text-to-Image Ultra – for maximum resolution and detail on individual hero images.
• Nano Banana Pro Edit – for object-level or style edits on existing assets.
• FLUX series – for cinematic, high-impact visuals and experimental aesthetics.
• Seedream series – for cost-effective, style-consistent illustration and multi-image set generation.
• Qwen Image series – for strong LoRA support, flexible style transfer, and advanced controllability.

Authentication

For authentication details, please refer to the Authentication Guide.

API Endpoints

Submit Task & Query Result

set -euo pipefail

export WAVESPEED_API_KEY="your-api-key"

REQUEST_BODY=$(cat <<'JSON'
{
  "prompt": "A cinematic ocean wave at sunrise, highly detailed",
  "aspect_ratio": "3:2",
  "num_images": 2,
  "output_format": "png"
}
JSON
)

# 1. Submit the prediction once. Never retry this request automatically:
# a retry after an unclear failure can create a second, billed task.
CHECK='The task may already exist. Check History or the predictions list before submitting again.'
if ! SUBMIT_RESPONSE=$(curl --silent --show-error --fail-with-body \
  --connect-timeout 10 --max-time 60 \
  -X POST "https://api.wavespeed.ai/api/v3/google/nano-banana-pro/text-to-image-multi" \
  -H "Authorization: Bearer ${WAVESPEED_API_KEY}" \
  -H "Content-Type: application/json" \
  -d "${REQUEST_BODY}"); then
  printf 'Submit failed. %s\n' "${CHECK}" >&2
  exit 1
fi

TASK=$(printf '%s' "${SUBMIT_RESPONSE}" | jq 'if type == "object" and has("data") then .data else . end')
PREDICTION_ID=$(printf '%s' "${TASK}" | jq -r '.id // empty')
if [ -z "${PREDICTION_ID}" ]; then
  printf 'Submission response did not contain a prediction id. %s\n' "${CHECK}" >&2
  exit 1
fi
RESULT_URL="https://api.wavespeed.ai/api/v3/predictions/${PREDICTION_ID}/result"

# 2. Poll until the prediction reaches a final status.
BODY_FILE=$(mktemp)
trap 'rm -f "${BODY_FILE}"' EXIT
while true; do
  # Transient errors (429, 5xx) are retried, honoring Retry-After.
  HTTP_CODE=$(curl --silent --show-error --fail-with-body \
    --connect-timeout 10 --max-time 60 --retry 5 --retry-delay 2 \
    --output "${BODY_FILE}" --write-out '%{http_code}' \
    "${RESULT_URL}" \
    -H "Authorization: Bearer ${WAVESPEED_API_KEY}" || true)
  case "${HTTP_CODE}" in
    2??) ;;
    000|429|5??)  # dropped connection or temporary error: keep polling
      printf 'Task %s: HTTP %s, polling again in 10s\n' "${PREDICTION_ID}" "${HTTP_CODE}" >&2
      sleep 10; continue ;;
    *) printf 'Task %s: HTTP %s\n' "${PREDICTION_ID}" "${HTTP_CODE}" >&2
      cat "${BODY_FILE}" >&2; exit 1 ;;
  esac
  RESULT=$(jq 'if type == "object" and has("data") then .data else . end' "${BODY_FILE}")
  STATUS=$(printf '%s' "${RESULT}" | jq -r '.status // empty')

  case "${STATUS}" in
    completed) printf '%s\n' "${RESULT}" | jq '.outputs'; break ;;
    failed|cancelled|timeout|deleted)
      printf 'Task %s ended with status %s\n' "${PREDICTION_ID}" "${STATUS}" >&2
      printf '%s\n' "${RESULT}" | jq . >&2; exit 1 ;;
    *) sleep 2 ;;
  esac
done

Parameters

Task Submission Parameters

Request Parameters

Every model also accepts the global request options enable_sync_mode and enable_base64_output; see their pages for details.

Response Parameters

Result Request Parameters

Result Response Parameters