OpenClaw Integration
Use the Creaa skill in OpenClaw, or call the same image and video endpoints directly with a single API key.
This page covers installation, API authentication, example requests, supported models, and billing entry points.
Seedance 2.0 is the default recommendation for most video tasks
For common OpenClaw and API video workflows, Seedance 2.0 is the most practical default because it is fast, stable, and currently priced below the official channel.
Quick Start
Get up and running in 3 steps.
Install the Skill
Create the skill folder and save the SKILL.md file.
mkdir -p ~/.openclaw/skills/creaa-ai
# Download skill file
curl -sL https://creaa.ai/api/open/v1/skill -o ~/.openclaw/skills/creaa-ai/SKILL.md
Set Your API Key
Configure your API key as an environment variable.
openclaw env set CREAA_API_KEY your_api_key_here
Usage
Once installed, just ask your agent to generate images or videos.
"Generate a photo of a sunset over the ocean using seedream-5.0"
"Create a 5-second video of a cat walking on the beach"
"Generate 4 product photos in 16:9 ratio with z-image-turbo"
API Quickstart
Use the same API key with any HTTP client. Submit a task, then poll until the result is ready.
Base URL & Auth
Authenticate with Bearer auth using your Creaa API key.
export CREAA_API_BASE=https://creaa.ai/api/open/v1
# API key
export CREAA_API_KEY=your_api_key_here
# List available models
curl -s "$CREAA_API_BASE/models" \
-H "Authorization: Bearer $CREAA_API_KEY"
Generate An Image
Call the image endpoint with a prompt, model, ratio, and count.
-X POST \
-H "Authorization: Bearer $CREAA_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"prompt": "Minimal product photo of a glass perfume bottle on warm stone, luxury commercial lighting",
"model": "seedream-5.0",
"aspect_ratio": "9:16",
"n": 1
}'
Generate A Video
Supports text_to_video plus Seedance 2 multi-modal image_to_video workflows. Submit first, then poll the task endpoint.
Seedance 2 keeps the same semantics as the Ark official SDK: use omni_reference for reference-image workflows, use first_last_frames for start/end frame control. The API also accepts reference_video_urls and reference_audio_urls. Audio must be combined with image or video references; audio-only and text+audio are not supported.
- text_to_video: no image input, prompt only.
- image_to_video + omni_reference: one main image plus extra reference images. Best when you want the model to keep identity, styling, props, or composition cues.
- image_to_video + first_last_frames: provide both start and end frames. Best when you want a controlled transition from frame A to frame B.
- Multi-modal references: you can also pass
reference_video_urlsandreference_audio_urlstogether with image/video references. Audio cannot be used by itself, and text + audio alone is rejected. - Compatibility: if your existing code uses
reference_images_urls,video_input_mode,start_frame_url, orend_frame_url, Creaa accepts those aliases directly.
curl -s "$CREAA_API_BASE/videos/generate" \
-X POST \
-H "Authorization: Bearer $CREAA_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"prompt": "A stylish woman unboxes a skincare serum in a bright bathroom, cinematic UGC style",
"model": "seedance-2.0",
"mode": "text_to_video",
"duration": 5,
"aspect_ratio": "9:16"
}'
curl -s "$CREAA_API_BASE/videos/generate" \
-X POST \
-H "Authorization: Bearer $CREAA_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"prompt": "Luxury perfume bottle rotates on a marble pedestal, soft cinematic light and floating mist",
"model": "seedance-2.0",
"mode": "image_to_video",
"image_mode": "omni_reference",
"image_url": "https://example.com/perfume-main.jpg",
"reference_image_urls": [
"https://example.com/perfume-detail-1.jpg",
"https://example.com/perfume-detail-2.jpg"
],
"aspect_ratio": "9:16",
"duration": 8
}'
curl -s "$CREAA_API_BASE/videos/generate" \
-X POST \
-H "Authorization: Bearer $CREAA_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"prompt": "A fashion model turns toward the camera and speaks in sync with the reference audio, luxury ad lighting",
"model": "seedance-2.0",
"mode": "image_to_video",
"image_mode": "omni_reference",
"image_url": "https://example.com/model-main.jpg",
"reference_audio_urls": [
"https://example.com/voice-guide.mp3"
],
"aspect_ratio": "9:16",
"duration": 5
}'
curl -s "$CREAA_API_BASE/videos/generate" \
-X POST \
-H "Authorization: Bearer $CREAA_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"prompt": "A futuristic sports car accelerates out of a tunnel into sunrise city light",
"model": "seedance-2.0",
"mode": "image_to_video",
"image_mode": "first_last_frames",
"first_frame_url": "https://example.com/car-start.jpg",
"last_frame_url": "https://example.com/car-end.jpg",
"aspect_ratio": "16:9",
"duration": 5
}'
curl -s "$CREAA_API_BASE/tasks/task_id_here" \
-H "Authorization: Bearer $CREAA_API_KEY"
# Optional: check API key usage
curl -s "$CREAA_API_BASE/usage" \
-H "Authorization: Bearer $CREAA_API_KEY"
Available Models
13 models for image and video generation, from budget-friendly to premium quality.
Why Creaa API
13+ Models, One API
Access Veo, Sora, Seedance, Kling, Runway and more through a single unified API.
Simple & Fast
Submit a prompt, get a task ID, poll for results. Images in ~10s, videos in 1-5 min.
Pay Per Use
No subscriptions required. Buy credits and use them across any model, anytime.
Developer Friendly
RESTful API with Bearer auth. Works with OpenClaw, curl, or any HTTP client.
Global CDN
Generated files served via Cloudflare CDN for fast downloads worldwide.
Secure
API keys with rate limiting, usage tracking, and per-key credit isolation.
Create your API key first
Start from your profile page. Create an API key first, then recharge credits to that key when you are ready.