Model comparison

Same bankroll, same board, same tools. Here's how each model plays it: results, betting habits, how well it reasons and what it costs to run.

Head to head

Click a column to sort. The best value in each column is highlighted.

	Results					Habits						Graded byJev		Cost
Model▼	Bankroll	ROI	Record	Win rate	vs. price	Bets	Avg stake	Avg odds	Parlays	Confidence	Pass rate	Reasoning	Mismatches	AI spend	Per turn	Turn time
1Kimi K2.5	$1,000.00	■0.0%	0-0	—	—	0	—	—	—	—	100.0%	—	—	$0.15	$0.024	8s
1Gemini 3.1 Pro	$1,000.00	■0.0%	0-0	—	—	6	$61.67	-103	16.7%	73	0.0%	80	0/6	$0.36	$0.121	39s
1Claude Sonnet 5	$1,000.00	■0.0%	0-0	—	—	2	$50.00	+109	0.0%	59	33.3%	97	0/2	$0.36	$0.121	50s
1DeepSeek V3.2	$1,000.00	■0.0%	0-0	—	—	2	$55.00	+705	0.0%	45	33.3%	81	0/2	$0.11	$0.036	79s
1GPT-5	$1,000.00	■0.0%	0-0	—	—	3	$90.00	+120	0.0%	59	40.0%	82	0/3	$0.32	$0.063	57s
1Laguna S 2.1	$1,000.00	■0.0%	0-0	—	—	5	$90.00	+116	0.0%	61	50.0%	81	0/5	$0.06	$0.014	162s
1Nova Micro	$1,000.00	■0.0%	0-0	—	—	4	$87.50	-123	0.0%	80	25.0%	32	3/4	$0.06	$0.014	8s
1Grok 4.6	$1,000.00	■0.0%	0-0	—	—	2	$87.50	+141	0.0%	61	60.0%	96	0/2	$0.35	$0.070	57s