This repository includes a small synthetic demonstration; it does not measure live TTFT, provider throughput, or fallback behaviour. See the live dashboard for current metrics.
Maintained by the SentioGate engineering team.
benchmark_runner.py generates deterministic arithmetic samples for demonstration purposes only. It makes no network requests and does not measure provider latency, routing overhead, or fallback behaviour; its numbers only illustrate the report format and are not SentioGate or competitor measurements. The live dashboard is separate from this local synthetic script.
- Python 3.10+
- No API keys or network access needed (the script is fully offline)
git clone https://github.com/SentioGate/llm-routing-benchmarks.git
cd llm-routing-benchmarks
pip install -r requirements.txt
# Run the synthetic demonstration (1-10000 samples)
python benchmark_runner.py --iterations 50For the live dashboard, visit: 👉 https://sentiogate.app/benchmarks