Creaa AI

metadata: {"openclaw":{"requires":{"env":["CREAA_API_KEY"]},"primaryEnv":"CREAA_API_KEY"}}

Creaa AI - Image, Edit & Video Generation

Generate and edit AI images, and generate AI videos directly from your terminal using the Creaa API.

Get your API key at: https://creaa.ai/profile -> API Keys Top up API credits at: https://creaa.ai/pricing

Available Models

Image Models

Aspect ratio options per model:

• seedream-4.5: 1:1, 16:9, 9:16, 4:3, 3:4
• nano-banana-2 / nano-banana-pro: auto, 1:1, 9:16, 16:9, 3:2, 2:3
• gpt-image-1.5: 1:1, 3:2, 2:3
• z-image-turbo: 1:1, 9:16, 16:9

Video Models

Aspect ratio options per model:

• veo-3.1: 16:9, 9:16
• seedance-2.0: 9:16, 16:9, 1:1, 3:4, 4:3
• sora-2-pro: 9:16, 16:9
• seedance-1.5-pro: 9:16, 16:9, 1:1, 3:4, 4:3
• kling-3.0: 16:9, 9:16
• hailuo-2.3: 9:16, 16:9
• runway-gen-4.5: 16:9, 9:16, 1:1, 4:3, 3:4, 21:9
• sora-2: 9:16, 16:9

Available Commands

All generation and edit tasks are asynchronous: submit a task, get a task_id, then poll for results.

Generate Image

Step 1: Submit task

curl -s -X POST "https://creaa.ai/api/open/v1/images/generate" \
  -H "Authorization: Bearer $CREAA_API_KEY" \
  -H "Content-Type: application/json" \
  -H "X-Source: openclaw" \
  -d '{
    "prompt": "<user prompt here>",
    "model": "<model_id>",
    "aspect_ratio": "<ratio>",
    "n": <count>
  }'

Parameters:

• prompt (required): Image description
• model (optional): Model ID from the image models table above. Default: seedream-4.5
• aspect_ratio (optional): Aspect ratio like 1:1, 16:9, 9:16. Must match the model's supported ratios. Default varies by model.
• n (optional): Number of images, 1-4. Default 1

Response:

{
  "success": true,
  "task_id": "abc123-def456",
  "task_type": "image",
  "model": "seedream-4.5"
}

Step 2: Poll status (wait 5-10 seconds between polls for images)

curl -s "https://creaa.ai/api/open/v1/tasks/<task_id>" \
  -H "Authorization: Bearer $CREAA_API_KEY" \
  -H "X-Source: openclaw"

Status Response (completed):

{
  "success": true,
  "task_id": "abc123-def456",
  "status": "completed",
  "progress": 1.0,
  "result_url": "https://...",
  "result_urls": ["https://...", "https://..."]
}

Image generation typically takes 10-60 seconds. Poll every 5 seconds.

Edit Image

Use this endpoint for:

• Single-image edit: image_url
• Multi-image reference edit / composition: image_urls

Step 1: Submit task

curl -s -X POST "https://creaa.ai/api/open/v1/images/edit" \
  -H "Authorization: Bearer $CREAA_API_KEY" \
  -H "Content-Type: application/json" \
  -H "X-Source: openclaw" \
  -d '{
    "prompt": "<edit prompt here>",
    "image_url": "https://...",
    "model": "<model_id>",
    "aspect_ratio": "<ratio>",
    "n": <count>
  }'

Single-image edit parameters:

• prompt (required): Edit instruction
• image_url (required for single-image edit): Public source image URL
• model (optional): Model ID from the image models table above. Default: seedream-4.5
• aspect_ratio (optional): Must match the model's supported ratios
• n (optional): Number of output images, 1-4. Default 1

Multi-image edit / composition example:

curl -s -X POST "https://creaa.ai/api/open/v1/images/edit" \
  -H "Authorization: Bearer $CREAA_API_KEY" \
  -H "Content-Type: application/json" \
  -H "X-Source: openclaw" \
  -d '{
    "prompt": "<combine these references into one result>",
    "image_urls": ["https://...", "https://..."],
    "model": "nano-banana-2",
    "aspect_ratio": "16:9",
    "n": 1
  }'

Multi-image edit notes:

• image_urls: 2-N public image URLs, where N depends on the model
• Max reference images by model:
  • seedream-4.5: up to 4 input images
  • nano-banana-2: up to 14 input images
  • nano-banana-pro: up to 10 input images
  • gpt-image-1.5: up to 10 input images
• Only models with multi_image_edit capability support this mode
• Only public http(s) URLs are accepted

Response:

{
  "success": true,
  "task_id": "abc123-def456",
  "task_type": "image_edit",
  "model": "seedream-4.5"
}

Step 2: Poll status

curl -s "https://creaa.ai/api/open/v1/tasks/<task_id>" \
  -H "Authorization: Bearer $CREAA_API_KEY" \
  -H "X-Source: openclaw"

Status Response (completed):

{
  "success": true,
  "task_id": "abc123-def456",
  "status": "completed",
  "progress": 1.0,
  "result_url": "https://...",
  "result_urls": ["https://...", "https://..."]
}

Image editing typically takes 10-90 seconds. Poll every 5 seconds.

Generate Video

Step 1: Submit task

curl -s -X POST "https://creaa.ai/api/open/v1/videos/generate" \
  -H "Authorization: Bearer $CREAA_API_KEY" \
  -H "Content-Type: application/json" \
  -H "X-Source: openclaw" \
  -d '{
    "prompt": "<user prompt here>",
    "model": "<model_id>",
    "mode": "<text_to_video|image_to_video>",
    "duration": <seconds>,
    "aspect_ratio": "<ratio>"
  }'

Parameters:

• prompt (required): Video description
• model (optional): Model ID from the video models table above. Default: veo-3.1
• mode (optional): text_to_video (default) or image_to_video
• image_url (required for image_to_video): URL of the source image
• duration (optional): Duration in seconds, must match the model's supported durations. Default varies by model.
• aspect_ratio (optional): e.g. 16:9, 9:16, 1:1. Must match the model's supported ratios.

Response:

{
  "success": true,
  "task_id": "abc123-def456",
  "task_type": "video",
  "model": "veo-3.1"
}

Step 2: Poll status (wait 15 seconds between polls for videos)

curl -s "https://creaa.ai/api/open/v1/tasks/<task_id>" \
  -H "Authorization: Bearer $CREAA_API_KEY" \
  -H "X-Source: openclaw"

Status Response (completed):

{
  "success": true,
  "task_id": "abc123-def456",
  "status": "completed",
  "progress": 1.0,
  "result_url": "https://..."
}

Status values: pending -> processing -> completed or failed. Video generation typically takes 60-300 seconds. Poll every 15 seconds.

Unified Task Status Endpoint

Image generation, image edit, and video tasks all use the same status endpoint:

curl -s "https://creaa.ai/api/open/v1/tasks/<task_id>" \
  -H "Authorization: Bearer $CREAA_API_KEY" \
  -H "X-Source: openclaw"

Keep polling until status is completed or failed.

• For images: poll every 5 seconds
• For videos: poll every 15 seconds

List Available Models

curl -s "https://creaa.ai/api/open/v1/models" \
  -H "Authorization: Bearer $CREAA_API_KEY" \
  -H "X-Source: openclaw"

Returns full model catalog with pricing, supported aspect ratios/durations, capabilities, and max_upload_images for image-edit models.

Check Usage

curl -s "https://creaa.ai/api/open/v1/usage" \
  -H "Authorization: Bearer $CREAA_API_KEY" \
  -H "X-Source: openclaw"

Usage is tracked against the API key's own credit pool (api_credits), not the website account balance.

Billing Notes

• Open API requests charge the API key credit pool, not the user's main website balance.
• If task submission fails before the job is queued, credits are refunded immediately.
• If a queued task fails, credits are refunded automatically.
• For image generation and image editing, partial success is billed by successful outputs; unused credits are refunded automatically.
• For video generation, successful jobs keep the full charge; failed jobs are refunded in full.

Model Selection Tips

Image models:

• seedream-4.5 - Best quality, great default choice
• nano-banana-2 - Best default for multi-image editing / composition
• nano-banana-pro - Higher quality image editing
• gpt-image-1.5 - Strong image editing option via Replicate-backed path
• z-image-turbo - Cheapest at 1 credit, fastest generation

Video models:

• veo-3.1 - Highest quality, 8s fixed duration
• sora-2 - Most affordable at 10 credits/sec
• sora-2-pro - Good balance of quality and cost
• runway-gen-4.5 - Image-to-video only, high quality

Error Handling

If any API call returns "success": false, display the error message to the user. Common errors:

• invalid_api_key: The API key is missing or wrong. User should check their key at https://creaa.ai/profile
• rate_limit_exceeded: Too many requests per minute. Wait and retry.
• Insufficient credits: Not enough credits. User should top up at https://creaa.ai/pricing
• Only public http(s) image URLs are allowed: The image URL points to localhost, private network, or a non-http(s) source.
• Model <id> does not support multi_image_edit: The requested model can edit a single image but cannot do multi-image composition.