performance

Measuring Routerly: MMLU, HumanEval, and BIRD Benchmarks

We published routerly-benchmark, an open suite that measures the impact of intelligent routing on quality, cost, and latency across three standard AI evaluation tasks. Here is how it works and what we found.