02 — Run · Round 1 · Round robin · rr-sonnet5-gpt56terra-g3
Claude Sonnet 5 vs GPT-5.6 Terra
GPT-5.6 Terra destroyed the enemy base after 5.3 game-minutes · 8.2 min wall clock
LOSS russia · west
Claude Sonnet 5
anthropic/claude-sonnet-5
Value destroyed$200Value lost$11,850
Units killed / lost2 / 3Buildings killed / lost0 / 12
Peak army$300Orders issued45
Decision turns (failed)40 (0)Mean latency4.7sModel cost$0.242
WIN germany · east
GPT-5.6 Terra
openai/gpt-5.6-terra
Value destroyed$11,850Value lost$200
Units killed / lost3 / 2Buildings killed / lost12 / 0
Peak army$2,700Orders issued42
Decision turns (failed)40 (0)Mean latency3.3sModel cost$0.287
Evaluation
passchecks 7/7 verdict computed from the gating checks
Round-1 run: written before the runbook moved to evals v2, so these are the older self-reported checks.
| Status | Kind | Check | Detail |
|---|---|---|---|
| pass | code_check | result-json | result json |
| pass | code_check | has-winner | has winner |
| pass | code_check | decisions-logged | decisions logged |
| pass | code_check | replay-saved | replay saved |
| pass | code_check | video-rendered | video rendered |
| pass | code_check | no-llm-outage | no llm outage |
| pass | code_check | video-complete | video complete |
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.