Skip to main content
Execute your serverless endpoint synchronously with input data and receive the results immediately. This is the primary way to run inference on your deployed AI models.

​
Path Parameters

  • endpointId: The unique identifier of the endpoint to execute

​
Request Body

The request body should contain an input object with your model-specific data: 
{
    "input": {
        "prompt": "Explain quantum computing in simple terms",
        "max_tokens": 100,
        "temperature": 0.7
    }
}

​
Example Usage

​
Text Generation

curl -X POST "https://api.tensorone.ai/v2/endpoints/ep_text_model/runsync" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "input": {
      "prompt": "Write a haiku about AI",
      "max_tokens": 50
    }
  }'

​
Image Generation

curl -X POST "https://api.tensorone.ai/v2/endpoints/ep_image_model/runsync" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "input": {
      "prompt": "A serene mountain landscape at sunset",
      "width": 512,
      "height": 512,
      "steps": 30
    }
  }'

​
Response

Returns the execution results: 
{
    "output": "Artificial minds learn,\nProcessing data with care,\nFuture unfolds bright.",
    "executionTime": 1.23,
    "status": "completed",
    "metadata": {
        "model": "llama-2-7b",
        "tokensUsed": 47,
        "cost": 0.0023
    }
}

​
Response Fields

  • output: The result from your model (format varies by model type)
  • executionTime: Time taken to execute in seconds
  • status: Execution status (completed or failed)
  • metadata: Additional information about the execution

​
Error Handling

Common error responses:

​
400 Bad Request

{
    "error": "INVALID_INPUT",
    "message": "Missing required field: prompt",
    "details": {
        "field": "input.prompt",
        "reason": "This field is required"
    }
}

​
404 Not Found

{
    "error": "ENDPOINT_NOT_FOUND",
    "message": "Endpoint ep_invalid does not exist"
}

​
500 Internal Server Error

{
    "error": "EXECUTION_FAILED",
    "message": "Model execution failed",
    "details": {
        "reason": "Out of memory error"
    }
}

​
Best Practices

  • Input Validation: Always validate your input data before sending requests
  • Error Handling: Implement retry logic for transient failures
  • Timeouts: Set appropriate request timeouts (default: 300 seconds)
  • Rate Limiting: Respect rate limits to avoid throttling
For long-running tasks, consider using asynchronous execution or streaming responses through our SDK libraries.
Execution costs are charged per request. Monitor your usage to avoid unexpected bills.