Sierra Open-Sources Hyper-τ-Bench, a Benchmark for Agent Construction

Sierra has open-sourced hyper-τ-bench, a long-horizon benchmark designed to evaluate whether AI coding agents can successfully construct working…