Google Gemini 3 Pro Image Edit API Documentation

Google Gemini 3 Pro Image Edit API Documentation

Playground

Google’s Gemini 3.0 Pro (Gemini 3.0 Pro Image) Edit enables image editing and text-to-image generation with 4K-capable output for mobile devices. Ready-to-use REST inference API, best performance, no coldstarts, affordable pricing.

Features

Gemini 3.0 Pro Image Edit (Gemini 3.0 Pro Image) is Google’s advanced AI-powered image editing and generation model, designed to make visual transformation as intuitive as describing it in words. Built on Google’s cutting-edge computer vision and generative research, it combines precision, flexibility, and semantic awareness for professional-grade editing.

🌟 Why it stands out

• Natural Language Editing Modify images using simple text instructions — no masking, layering, or manual tools required.
• Context-Aware Understanding Accurately interprets scene structure, spatial relationships, and object semantics for realistic results.
• Style and Tone Preservation Keeps lighting, shadows, and texture consistent with the original image while applying changes seamlessly.
• High Precision Control Excels at fine-grained edits such as color adjustments, object replacement, or composition shifts with minimal distortion.
• Creative Versatility Suitable for concept art, photography, advertising design, and everyday content creation.

⚙️ How to use

• Input: existing image + text prompt

• Output: edited image (JPEG/PNG)

• Size: 1:1, 4:3, 16:9, 21:9, and so on.

• Supports style transfer, relighting, background replacement, and object modification

• Works with natural prompts like:

• “Replace the cloudy sky with a clear sunset.”

• “Add soft studio lighting and a modern background.”

• “Turn the model’s outfit into a formal business suit.”

💰 Pricing

💡 Best Use Cases

• Marketing & Branding — Update campaign visuals without reshooting.
• Product Photography — Adjust materials, lighting, or layout instantly.
• Social Media & Content Creation — Generate multiple variations with minimal effort.
• Artistic Design — Experiment with colors, styles, and compositions effortlessly.

📝 Notes

Please ensure your prompts comply with Google’s Safety Guidelines. If an error occurs, review your prompt for restricted content, adjust it, and try again.

🌏 Where gemini-3-pro-image Edit Fits In

Compare gemini-3-pro-image Edit with:

• FLUX.1 [dev] – Nano Banana Pro Edit focuses on semantic understanding and layout-aware editing via Gemini 3’s reasoning, making it ideal for complex, text-driven transformations without manual masking. FLUX.1 [dev] emphasizes maximum resolution control and fine detail preservation for highly technical workflows.

• Stable Diffusion 3.5 – Nano Banana Pro Edit leverages Google’s production-scale multimodal training and advanced reasoning for context-aware edits that understand object relationships and composition. Stable Diffusion 3.5 offers open-source flexibility for custom fine-tuning and local deployment in highly specialized pipelines.

• Gpt-Image-1 (OpenAI) – Nano Banana Pro Edit emphasizes layout control, multilingual on-image text, and tightly directed edits for design and marketing workflows, while openai/gpt-image-1 shines as a general-purpose creative generator with strong style variety and fast, natural-language image synthesis for broad consumer and developer use.

• Original Nano Banana (Gemini 2.5 Flash Image) – Nano Banana Pro Edit trades pure speed for quality, delivering better reasoning, sharper text, improved character consistency, and richer camera controls at a higher unit cost. Original Nano Banana remains ideal for rapid, low-latency iterations and lightweight edits.

• Seedream – Nano Banana Pro Edit is tuned for reliable typography, photo-real edits, and mixed media layouts, while SeeDream excels at fast, stylized T2I generation with strong anime and illustration aesthetics, making it a good choice for heavily stylized concept art.

• Qwen Image 2509 – Nano Banana Pro Edit focuses on high-fidelity 4K outputs and multilingual on-image design control, whereas Qwen Image shines in open-source ecosystems and document-style rendering, offering flexible integration for developer-centric and research workflows.

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",
  "images": [
    "https://interactive-examples.mdn.mozilla.net/media/cc0-images/painted-hand-298-332.jpg"
  ],
  "aspect_ratio": "1:1",
  "resolution": "1k",
  "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/gemini-3-pro-image/edit" \
  -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