Managing and Reducing Costs › Strategy 6: Use Prompt Compression

import os
from anthropic import Anthropic

client = Anthropic(api_key=os.getenv("ANTHROPIC_API_KEY"))

def summarize_context(long_context, max_length=500):
    """Compress long context into a shorter summary."""
    if len(long_context) <= max_length:
        return long_context
    
    # Use the model to create a concise summary
    response = client.messages.create(
        model="claude-sonnet-4-5",
        max_tokens=200,
        messages=[{
            "role": "user",
            "content": f"Summarize this in {max_length//4} words or less:\n\n{long_context}"
        }]
    )
    
    return response.content[0].text

def respond_with_context(query, long_context):
    """Answer query using compressed context."""
    # Compress the context first
    compressed_context = summarize_context(long_context, max_length=500)
    
    # Use compressed context in the actual query
    full_prompt = f"Context: {compressed_context}\n\nQuestion: {query}"
    
    response = client.messages.create(
        model="claude-sonnet-4-5",
        max_tokens=512,
        messages=[{"role": "user", "content": full_prompt}]
    )
    
    return response.content[0].text

## Example: Long document that needs to be included
long_document = """
[Imagine a 5000-word document about machine learning here...]
Machine learning is a field of artificial intelligence that focuses on...
[... many more paragraphs ...]
"""

## Without compression: ~5000 tokens input
## With compression: ~500 tokens input
## Savings: ~90% on input tokens for this context

query = "What are the key concepts in this document?"
answer = respond_with_context(query, long_document)
print(answer)