README.md

LLM Prompt Caching Benchmark

This Python script benchmarks the performance improvement of prompt caching for the OpenAI GPT-4o and Anthropic Claude sonnet-3.5 language models. It measures the latency for API calls with and without prompt caching by requesting a small number of output tokens, approximating the time to first byte. It then calculates the percentage improvement in latency.

Usage

1. Install the required Python packages:

pip install -r requirements.txt

2. Set the API keys for OpenAI and Anthropic in the benchmark.py script.
3. Run the script:

python benchmark.py