evals
Run evals on Jetty

02 — Run · Round 2 · Round robin · r2-rr-gpt56terra-gpt56luna-g3

GPT-5.6 Terra vs GPT-5.6 Luna

GPT-5.6 Terra destroyed the enemy base after 7.8 game-minutes · 10.2 min wall clock

WIN france · west
GPT-5.6 Terra
openai/gpt-5.6-terra
Value destroyed$13,700Value lost$7,300 Units killed / lost26 / 54Buildings killed / lost8 / 1 Peak army$4,900Orders issued97 Decision turns (failed)59 (0)Mean latency3.7sModel cost$0.451
LOSS ukraine · east
GPT-5.6 Luna
openai/gpt-5.6-luna
Value destroyed$7,300Value lost$13,700 Units killed / lost54 / 26Buildings killed / lost1 / 8 Peak army$2,050Orders issued89 Decision turns (failed)59 (0)Mean latency3.7sModel cost$0.048

Evaluation

passchecks 10/10 verdict computed from the gating checks

StatusKindCheckDetail
passstepenvironmentEnvironment check (snapshot, engine, content, key)
passsteplaunchLaunch run-match as a detached process
passstepsuperviseSupervise in the foreground until run-match exits
passsteprenderRender the 1080p recording
passstepnotesAppend Notes to summary.md
passcode_checkoutputs-existall present
passcode_checkgame-has-winnergpt56terra won: base destroyed
passcode_checkdecisions-loggedturns per slot: {'Multi0': 59, 'Multi1': 59}
passcode_checkllm-turns-validfailed/total per slot: {'Multi0': '0/59', 'Multi1': '0/59'}
passcode_checkvideo-decodesduration 80.5s
passcode_checkvideo-completevideo ends at tick 11774 of 11774
passcode_checklimits-respectedended after 10.2 wall-minutes (budget 30.0 min): base destroyed
passchecklistnotes-explain-winsummary.md notes section cites final thoughts of both models and explains Terra base-destruction victory
passchecklistplayers-match-requestresult.json records gpt56terra as Multi0 and gpt56luna as Multi1, matching the requested players
passchecklistno-silent-outagerun.log shows continuous decision turns with 0 llm_failures for both players across all 59 turns each

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.

Ready to stop guessing if your outputs are good?

Get Started Book a 15-minute walkthrough
Connect your agent to Jetty →