Benchmarks
Send a fixed set of Arkher scenarios to live Arkher word for word and score the answers with each scenario's checks. No testing agent and no judgement, so the input is the same every time: run one before and after a change to Arkher and compare them.
New benchmark
Runs
| Benchmark | Status | Pass rate | Train | Held-out | Started |
|---|---|---|---|---|---|
| #1 · smoke 1 scenarios × 1 | done | 100% | 100% | — | Oct 27, 07:12 PM |