Daily Arena

Match replay

Replaying a stored match — no models are called.

⚖ Multi-judge consensus — our trademark
Tokonomix multi-council + judge + blind-spot detection — lower cost, and it catches the mistakes one model misses.
Multi-council · lower costMulti-judge · cross-familyBlind-spot detection · catch the missed mistakeN-team · groups vs each other
Game type
Turns: 5
Speed1×
customer_service · roundTurn 0 / 5
The cheapest model that keeps up on quality appears here.
0 / 5
Council A
Mistral-7B-Instruct-v0.3 · Mistral-Nemo-Instruct-2407 · Mistral-Small-3.2-24B-Instruct-2506
CO
€—score —
HP
100
Council B
gpt-5 · gpt-4 · gpt-4-turbo
CO
€—score —
HP
100
Frontier C
Claude Opus 4.7
FR
€—score —
HP
100
Customer
Press “Next turn” to begin.

Final verdict — cost, quality & voorsprong

PlayersCostQualityWinsVoorsprong / status
Council A€0.124860.67058 HP
Council B€0.234752.67046 HP
Frontier C€0.0814865100 HP
0 / 5Drone damage = jury-majority strength · HP = live voorsprong · € = real cost

Honesty boundary

Advantage starts at 100; each turn the weakest active model loses the derived damage — damage = 16 + 24·margin, margin = (winner − runner-up) ÷ score-scale (deriveRoundOutcomes v8.1-tokonomix).

An exact tie has no decisive winner — no strike, no damage that turn.

Reaching 0 advantage is NOT elimination: every model still answers each turn. The real winner is the end-of-round judge panel below, shown for all models.

Damage reflects the relative gap between top scores, not absolute quality — winning a low-scoring turn deals the same as winning a high-scoring one.

Score-scale is the highest turn-score seen in this replay (0–10 or 0–100); one high turn can make the others look closer.

Zero model dispatch — pure render of the stored round. Switching the view changes the picture, never the numbers.
Back to the arena
↺ Start a new round