How WaveSpeedAI Works
Understand the WaveSpeedAI platform architecture and workflow.
Platform Overview
WaveSpeedAI is a unified API platform that provides access to 1,000+ AI models from multiple providers through a single interface.
Your Application → WaveSpeedAI API → AI Models → Generated ContentKey Concepts
Models
AI models are the core of WaveSpeedAI. Each model specializes in a specific task:
- Text-to-Image — Generate images from text descriptions
- Image-to-Video — Animate images into videos
- Text-to-Video — Create videos from text
- Audio Generation — Generate speech, music, and sounds
See What are Models for details.
Predictions
A prediction (also called a task) is a single request to generate content. Each prediction:
- Has a unique ID for tracking
- Goes through status stages (created → processing → completed)
- Contains outputs (URLs to generated content)
See What are Predictions for details.
Request Flow
1. Submit Task
Send your request with parameters to the model endpoint:
POST /api/v3/{model_id}
{
"prompt": "Your prompt here",
...parameters
}2. Task Processing
WaveSpeedAI:
- Validates your request
- Routes to the appropriate model
- Queues and processes the task
- Generates content
3. Retrieve Results
Poll the task status until completion:
GET /api/v3/predictions/{task-id}/resultWhen status is completed, the outputs array contains URLs to your generated content.
Webhooks (Optional)
Instead of polling, pass an HTTPS callback URL in the webhook query parameter:
POST /api/v3/{model_id}?webhook=https%3A%2F%2Fyour-server.com%2FwebhookSee How to Use Webhooks for details.
Content Lifecycle
- Task created — Your request is received
- Processing — AI model generates content
- Completed — Content available via URLs
- Temporarily available — Generated media URLs generally expire within 7 days
- Expired — Keep your own copy of outputs you need to retain
Rate Limits
Your account tier determines:
- How many requests per minute
- How many concurrent tasks
See Account Levels & Rate Limits for details.