WaveSpeed Python SDK

pip install wavespeed

from wavespeed import WaveSpeed
 
# Initialize the client with your API key (or set WAVESPEED_API_KEY environment variable)
client = WaveSpeed(api_key="YOUR_API_KEY")
 
# Generate an image and wait for the result
prediction = client.run(
    modelId="wavespeed-ai/flux-dev",
    input={
        "prompt": "A futuristic cityscape with flying cars and neon lights",
        "size": "1024*1024",
        "num_inference_steps": 28,
        "guidance_scale": 5.0,
        "num_images": 1,
        "seed": -1,
        "enable_safety_checker": True
    }
)
 
# Print the generated image URLs
for i, img_url in enumerate(prediction.outputs):
    print(f"Image {i+1}: {img_url}")

import asyncio
from wavespeed import WaveSpeed
 
async def generate_image():
    # Initialize the client with your API key
    client = WaveSpeed(api_key="YOUR_API_KEY")
    
    try:
        # Generate an image and wait for the result
        prediction = await client.async_run(
            modelId="wavespeed-ai/flux-dev",
            input={
                "prompt": "A futuristic cityscape with flying cars and neon lights",
                "size": "1024*1024",
                "num_inference_steps": 28,
                "guidance_scale": 5.0,
                "num_images": 1,
                "seed": -1,
                "enable_safety_checker": True
            }
        )
        
        # Print the generated image URLs
        for i, img_url in enumerate(prediction.outputs):
            print(f"Image {i+1}: {img_url}")
    finally:
        # Always close the client when done
        await client.close()
 
# Run the async function
asyncio.run(generate_image())

from wavespeed import WaveSpeed
 
# Initialize the client with your API key
client = WaveSpeed(api_key="YOUR_API_KEY")
 
# Create a prediction without waiting
prediction = client.create(
    modelId="wavespeed-ai/flux-dev",
    input={
        "prompt": "A futuristic cityscape with flying cars and neon lights",
        "size": "1024*1024",
        "num_inference_steps": 28,
        "guidance_scale": 5.0,
        "num_images": 1,
        "seed": -1,
        "enable_safety_checker": True
    }
)
 
print(f"Prediction created with ID: {prediction.id}")
print(f"Initial status: {prediction.status}")
 
# Instead of waiting, manually poll for completion
import time
 
# Poll for status until completed or failed
while prediction.status not in ["succeeded", "failed"]:
    print(f"Current status: {prediction.status}")
    time.sleep(1)  # Wait for 1 second before checking again
    prediction = prediction.reload()  # Reload the prediction status
 
print(f"Final status: {prediction.status}")
 
# Print the generated image URLs if successful
if prediction.status == "succeeded":
    for i, img_url in enumerate(prediction.outputs):
        print(f"Image {i+1}: {img_url}")
else:
    print(f"Error: {prediction.error}")

# Set your API key as an environment variable
export WAVESPEED_API_KEY="your_api_key_here"
 
# Run the example script
python examples/generate_image.py --prompt "A futuristic cityscape with flying cars and neon lights"

# Run the async example script
python examples/async_generate_image.py --prompt "A futuristic cityscape with flying cars and neon lights"

# Create a prediction and poll for status
python examples/create_generate_image.py --prompt "A futuristic cityscape with flying cars and neon lights"

--prompt TEXT         Text description of the desired image (required)
--strength FLOAT      How much to transform the input image (0.0 to 1.0)
--size WIDTHxHEIGHT   Image dimensions (default: 1024*1024)
--steps INT           Number of inference steps (default: 28)
--guidance FLOAT      How closely to follow the prompt (default: 5.0)
--num-images INT      Number of images to generate (default: 1)
--seed INT            Random seed (-1 for random)
--safety              Enable content safety filtering

WaveSpeed(api_key)

run(modelId, input, **kwargs) -> Prediction

async_run(modelId, input, **kwargs) -> Prediction

create(modelId, input, **kwargs) -> Prediction

async_create(modelId, input, **kwargs) -> Prediction

prediction.id                # Unique ID of the prediction
prediction.model             # Model ID used for the prediction
prediction.status            # Status of the prediction (processing, completed, failed)
prediction.input             # Input parameters used for the prediction
prediction.outputs           # List of output image URLs
prediction.urls.get          # URL to get the prediction status
prediction.has_nsfw_contents # List of booleans indicating if each image has NSFW content
prediction.created_at        # Creation timestamp
prediction.error             # Error message (if any)
prediction.executionTime     # Time taken to execute the prediction in milliseconds

prediction.wait() -> Prediction     # Wait for the prediction to complete
prediction.reload() -> Prediction   # Reload the prediction status
 
# Async versions
await prediction.async_wait() -> Prediction    # Wait for completion using non-blocking HTTP requests
await prediction.async_reload() -> Prediction  # Reload status using non-blocking HTTP requests