How to Use the WaveSpeedAI Python SDK

The WaveSpeedAI Python SDK provides a simple way to integrate AI image and video generation into your Python applications. This guide covers everything you need to get started.

Prerequisites

Before you begin, make sure you have:

Python 3.8+ installed on your machine
A WaveSpeedAI API key from wavespeed.ai/settings/api-keys

Installation

Install the SDK using pip:

pip install wavespeed

Setting Up Authentication

The SDK needs your API key to authenticate requests. You have two options:

Option 1: Environment Variable (Recommended)

Set the WAVESPEED_API_KEY environment variable:

export WAVESPEED_API_KEY="your-api-key-here"

Then use the SDK directly:

import wavespeed

output = wavespeed.run(
    "wavespeed-ai/z-image/turbo",
    {"prompt": "Cat"}
)

Option 2: Pass API Key Directly

Import the Client class and pass your API key to the constructor:

from wavespeed import Client

client = Client(api_key="your-api-key-here")

Generating Your First Image

Here’s a complete example that generates an image using Z-Image Turbo:

import wavespeed

output = wavespeed.run(
    "wavespeed-ai/z-image/turbo",
    {"prompt": "A serene mountain landscape at sunset with golden light"}
)

print(output["outputs"][0])  # URL to the generated image

The run() function handles the entire workflow: submitting the request, polling for completion, and returning the result.

Uploading Files

For workflows that require input images (like image-to-video), use the upload() function to get a URL that WaveSpeedAI can access:

import wavespeed

# Upload a local image file
image_url = wavespeed.upload("./my-image.png")

# Use the uploaded image for video generation
video = wavespeed.run(
    "wavespeed-ai/wan-2.1/image-to-video",
    {
        "image": image_url,
        "prompt": "Camera slowly zooms in while clouds move in the background"
    }
)

print(video["outputs"][0])  # URL to the generated video

Configuration Options

Client Options

Configure retry behavior when initializing the client:

from wavespeed import Client

client = Client(
    api_key="your-api-key",
    max_retries=3,            # Max retries for failed requests
    max_connection_retries=5, # Max retries for connection errors
    retry_interval=1.0        # Seconds between retries
)

Run Options

Configure individual run() calls:

import wavespeed

output = wavespeed.run(
    "wavespeed-ai/z-image/turbo",
    {"prompt": "A cute orange cat wearing a tiny hat"},
    timeout=60.0,           # Max seconds to wait for completion
    poll_interval=0.5,      # Seconds between status checks
    enable_sync_mode=True   # Use synchronous mode if available
)

Working with Different Models

Text-to-Image

Generate images from text descriptions:

import wavespeed

output = wavespeed.run(
    "wavespeed-ai/z-image/turbo",
    {
        "prompt": "A futuristic cityscape with flying cars and neon lights",
        "size": "1024x1024"
    }
)

Image-to-Video

Transform static images into videos:

import wavespeed

image_url = wavespeed.upload("./landscape.jpg")

video = wavespeed.run(
    "wavespeed-ai/wan-2.1/image-to-video",
    {
        "image": image_url,
        "prompt": "Gentle wind blowing through the trees"
    }
)

Text-to-Video

Generate videos directly from text:

import wavespeed

video = wavespeed.run(
    "wavespeed-ai/wan-2.1/t2v-480p",
    {"prompt": "A golden retriever running through a field of flowers"}
)

Async Support

The SDK supports async/await for non-blocking operations:

import asyncio
import wavespeed

async def generate_image():
    output = await wavespeed.async_run(
        "wavespeed-ai/z-image/turbo",
        {"prompt": "An astronaut riding a horse on Mars"}
    )
    return output["outputs"][0]

# Run the async function
image_url = asyncio.run(generate_image())
print(image_url)

Error Handling

For production applications, configure retries and handle errors gracefully:

from wavespeed import Client

client = Client(
    max_retries=3,
    max_connection_retries=5,
    retry_interval=1.0
)

try:
    output = client.run(
        "wavespeed-ai/z-image/turbo",
        {"prompt": "A beautiful sunset over the ocean"},
        timeout=120.0
    )
    print("Generated:", output["outputs"][0])
except Exception as e:
    print(f"Generation failed: {e}")