AI infrastructure for production
Test your AI workflow before you pay for it. Design, integrate, and validate everything
before connecting a provider.
- Free forever up to 25,000 requests/month
- No credit card required
- Free testing mode
$ curl https://api.modelriver.com/v1/ai/async \
-H "Authorization: Bearer YOUR_API_KEY" \
-d '{
"workflow": "book_review",
"messages": [{"role": "user", "content": "Hello!"}],
"user_id": "user_12345",
"task_id": "task_67890"
}'Real-time request processing
Watch requests flow through with
intelligent routing and automatic failover.
Toggle to dark mode

Live Logs
Waiting for requests...
Live Logs
Waiting for requests...
Everything you need to ship AI
without breaking things.

Auto failover if one model fails, we switch automatically.
Learn about failover
Stream AI responses live.
Explore streaming
Analytics track requests and speed easily.
See analytics
Debugging see every request and response.
Explore debugging
Workflows chain multiple AI steps together.
Explore webhooks
Structured output get clean, predictable JSON.
Explore structured outputUnified API Gateway
Simple integration with one endpoint and one API key
switch providers without changing your code.
REQUEST
{
"workflow": "my_book_task",
"messages": [
{
"role": "user",
"content": "Hello!"
}
],
"user_id": "user_12345",
"task_id": "task_67890"
}IN PROGRESS
{
"message": "success",
"status": "pending",
"channel_id": "a1b2c3d4-...",
"ws_token": "one-time-websocket-token",
"websocket_url": "wss://api.modelriver.com/socket",
"websocket_channel": "ai_response:PROJECT_ID:a1b2c3d4-..."
}SUCCESS
{
"message": "success",
"status": "success",
"data": {
"response": "Hello! How can I help?",
"intent": "greeting",
"confidence": 0.95
},
"model": "gpt-5.2",
"customer_data": {
"user_id": "user_12345",
"task_id": "task_67890"
},
"meta": {
"duration_ms": 1250,
"usage": {
"prompt_tokens": 21,
"completion_tokens": 42,
"total_tokens": 63
}
}
}Integration testing
Test before you ship with a free playground and catch issues.
Safe testing mode runs like production without affecting live users.
Free playground offers unlimited test requests at no extra cost.
Validate outputs lets you verify response format before going live.
Testing mode
Same settings as production, safe sandbox.
mode "testing"
structure "todo_struct"
failover enabled
rate_limits 30-300/min per IP,
40-1,000/min per project
40-1,000/min per project
cost free
Response preview
See exactly what your app will receive.
// sample response data
{
"data": {
"todo": {
"id": 1,
"title": "Finish project proposal",
"description": "Complete the first draft and review key points",
"completed": false,
"priority": "high"
}
}
}{
"data": {
"todo": {
"id": "integer",
"title": "string",
"description": "string",
"completed": "boolean",
"priority": "enum"
}
}
} Free playground testing
Ready to test your workflow
without burning credits?
Start free. 25k requests included.