Response Caching

Cache model responses to reduce API calls and costs.

Response caching allows you to cache model responses, which can significantly improve response times and reduce API costs during development and testing.

Note

For a detailed overview of response caching, see Response Caching.

Note

This is different from Anthropic's prompt caching feature. Response caching caches the entire model response, while prompt caching caches the system prompt to reduce processing time.

Basic Usage

Enable caching by setting cache_response=True when initializing the model. The first call will hit the API and cache the response, while subsequent identical calls will return the cached result.

1import time
2
3from kern.agent import Agent
4from kern.models.anthropic import Claude
5
6agent = Agent(model=Claude(id="claude-sonnet-4-5", cache_response=True))
7
8# Run the same query twice to demonstrate caching
9for i in range(1, 3):
10    print(f"\n{'=' * 60}")
11    print(
12        f"Run {i}: {'Cache Miss (First Request)' if i == 1 else 'Cache Hit (Cached Response)'}"
13    )
14    print(f"{'=' * 60}\n")
15
16    response = agent.run(
17        "Write me a short story about a cat that can talk and solve problems."
18    )
19    print(response.content)
20    print(f"\n Elapsed time: {response.metrics.duration:.3f}s")
21
22    # Small delay between iterations for clarity
23    if i == 1:
24        time.sleep(0.5)

Usage

Set up your virtual environment

1uv venv --python 3.12
2source .venv/bin/activate

1uv venv --python 3.12
2.venv\Scripts\activate

Set your API key

1export ANTHROPIC_API_KEY=xxx

Install dependencies

1uv pip install -U anthropic kern-ai

Run Agent

1python cache_model_response.py