evals
Run evals on Jetty
Pilot cup · cup1-final-g3

Jev Router vs Claude Sonnet 5

Jev Router won on score at the time limit (38,100 to 32,450) after 14.0 game-minutes · 25.6 min wall clock

Pilot game (run locally, before the Jetty runs) Replay (.orarep) Decision log Runtime config
WIN ukraine · west
Jev Router
typesafe/jev-router
Value destroyed$23,100Value lost$13,850 Units killed / lost31 / 53Buildings killed / lost1 / 4 Peak army$2,850Orders issued155 Decision turns (failed)106 (2)Mean latency9.7sModel cost$0.004
LOSS france · east
Claude Sonnet 5
anthropic/claude-sonnet-5
Value destroyed$13,850Value lost$23,100 Units killed / lost53 / 31Buildings killed / lost4 / 1 Peak army$5,700Orders issued197 Decision turns (failed)106 (0)Mean latency5.0sModel cost$0.849

Decision timeline

What each model saw as its situation and what it ordered, every 8 game-seconds. Click a turn to jump the video there.

Want this kind of eval for your own agents?

Every run on this site is a Jetty runbook: versioned instructions, a pinned sandbox, and a trajectory you can inspect. Point Jetty at your task and get the same receipts.