Skip to main content

​
Routing Profiles

PureRouter is a completely independent product from PureCPP. You can use PureRouter without needing PureCPP and vice versa.
PureRouter offers three main routing profiles that allow you to balance cost, quality, and speed according to your specific needs.

​
Available Profiles

The Economy profile is cost-optimized, directing your queries to more economical models that can still provide adequate responses for simple tasks.Ideal for:
  • Simple and direct queries
  • Basic text processing tasks
  • Scenarios where budget is a primary concern
  • Applications with high query volume
# Example using the Economy profile
response = client.router.infer(InferRequest(
    prompt="What is the capital of Brazil?",
    profile="economy"
))
The Balanced profile offers a balance between cost and quality, being the default option for most use cases.Ideal for:
  • General and everyday use
  • Medium complexity tasks
  • When you need reasonable quality responses without excessive costs
  • Applications with varied needs
# Example using the Balanced profile
response = client.router.infer(InferRequest(
    prompt="Explain how photosynthesis works in simple terms.",
    profile="balanced"
))
The Quality profile prioritizes response quality, directing your queries to the most advanced models available.Ideal for:
  • Complex and specialized tasks
  • Creative content generation
  • Advanced reasoning and problem solving
  • When precision and quality are critical
# Example using the Quality profile
response = client.router.infer(InferRequest(
    prompt="Write a detailed essay on the ethical implications of artificial intelligence in modern society.",
    profile="quality"
))

​
Profile Comparison

FeatureEconomyBalancedQuality
CostLowMediumHigh
QualityBasicGoodExcellent
Supported complexityLowMediumHigh
CreativityLimitedModerateHigh
ReasoningBasicGoodAdvanced
Ideal useSimple tasks, high volumeGeneral useComplex, specialized tasks

​
Choosing the Right Profile

  1. Evaluate task complexity - Simple tasks can use Economy, while complex tasks need Quality
  2. Consider your budget - If cost is a primary concern, start with Economy and scale as needed
  3. Test different profiles - Compare results to find the best balance for your specific use case
  4. Analyze query volume - For high volume, Economy may be more suitable; for low volume critical queries, Quality may be justified

​
Practical Examples

​
Example 1: Customer Service Chatbot

def customer_service_bot(query, importance):
    # Use Economy for simple FAQ questions
    if is_faq(query):
        return client.router.infer(InferRequest(
            prompt=query,
            profile="economy"
        ))
    
    # Use Balanced for most queries
    elif importance == "normal":
        return client.router.infer(InferRequest(
            prompt=query,
            profile="balanced"
        ))
    
    # Use Quality for VIP customers or critical issues
    elif importance == "high":
        return client.router.infer(InferRequest(
            prompt=query,
            profile="quality"
        ))

​
Example 2: Writing Assistant

def writing_assistant(task_type, content):
    if task_type == "grammar_check":
        # Simple grammar corrections can use Economy
        return client.router.infer(InferRequest(
            prompt=f"Correct the grammatical errors in the following text: {content}",
            profile="economy"
        ))
    
    elif task_type == "rewrite":
        # Text rewrites use Balanced
        return client.router.infer(InferRequest(
            prompt=f"Rewrite the following text to make it clearer and more concise: {content}",
            profile="balanced"
        ))
    
    elif task_type == "creative_content":
        # Creative content uses Quality
        return client.router.infer(InferRequest(
            prompt=f"Create creative and engaging content about: {content}",
            profile="quality"
        ))

​
Next Steps

Python SDK

Learn to use the PureRouter Python SDK

Deployments

Learn how to use specific deployments when needed

API Reference

Check the complete API documentation