🎮 Lmgame Bench: Leaderboard 🎲

🎮 Welcome to LMGame Bench!

We invite developers to implement their own gaming agents by replacing our baseAgent in customer_runner.py and evaluate them on our comprehensive benchmark. Visit our repository at https://github.com/lmgame-org/GamingAgent to get started and join the competition to see how your agent performs!

📊 Data Visualization

Number of Top Models to Display in All Views (max: 14)

1 14

Comparative Analysis (Radar Chart)

💡 Click a legend entry to isolate that model. Double-click additional ones to add them for comparison.

🕹️ Game Selection

⏰ Time Tracker

03/25/2025

🔄 Controls

📋 Detailed Results

💡 The slider above controls how many top models are shown in the radar chart, bar chart, and data table.

All data analysis can be replicated by checking this Jupyter notebook

	Player	Organization	Avg Normalized Score	Super Mario Bros Score	Sokoban Score	2048 Score	Candy Crush Score	Tetris Score	Ace Attorney Score
llama-4-maverick-17b-128e-instruct-fp8	anthropic	83.51	1025.3	1.33	1882.67	557.67	13.67	1.33

	Player	Organization	Avg Normalized Score	Super Mario Bros Score	Sokoban Score	2048 Score	Candy Crush Score	Tetris Score	Ace Attorney Score
1	🥇 o3-2025-04-16	openai	83.51	1955	2	7220	106	31	8
2	🥈 o1-2024-12-17	openai	40.76	1434	0	7176	90	13	3
3	🥉 gemini-2.5-pro-preview-05-06	google	39.78	1025.3	1	120.5	177.3	12.3	8
4	o4-mini-2025-04-16	openai	39.6	1348.3	1.33	1882.67	110.7	15	2
5	claude-sonnet-4-20250514	anthropic	37.79	n/a	0	3844	557.67	13.67	1.33
6	gemini-2.5-pro-preview-06-05	google	36.2	n/a	0.33	2232	496	13.67	1.33
7	gemini-2.5-flash-preview-04-17	google	33.04	1540.7	0	1738.67	97.7	19	1
8	claude-3-7-sonnet-20250219	anthropic	32.22	1430	0	2624	126.3	13	3
9	gemini-2.5-flash-preview-05-20	google	31.47	n/a	0	2750	254	16	2.33
10	gpt-4.1-2025-04-14	openai	29.83	1991.3	0	1113.33	101	13	0
11	claude-3-5-sonnet-20241022	anthropic	20.49	1540	0	84	17	12.3	1
12	gpt-4o-2024-11-20	openai	15.88	1028.3	0	176	59	14.7	0
13	random (x30)	unknown	14.91	986.97	0	1228	116.5	10.2	0
14	llama-4-maverick-17b-128e-instruct-fp8	meta	10.35	786	0	28	32.3	11.7	0

Note: 'n/a' in the table indicates no data point for that model.

	Player	Organization	Avg Normalized Score	Super Mario Bros Score	Sokoban Score	2048 Score	Candy Crush Score	Tetris Score	Ace Attorney Score
1	🥇 🎮 o3-2025-04-16 (GamingAgent)	openai	98.33	3445	8	7120	647	42	16
2	🥈 🎮 o1-2024-12-17 (GamingAgent)	openai	54.57	855	2.33	7580	159	35	16
3	🥉 🎮 o4-mini-2025-04-16 (GamingAgent)	openai	50.91	1448	5.33	4432	487.3	25.3	4
4	🎮 gemini-2.5-pro-preview-05-06 (GamingAgent)	google	46.66	1498.3	4.33	3586.67	416.3	23.3	7
5	🎮 deepseek-r1-0528 (GamingAgent)	deepseek	40.99	n/a	4.67	3330	491.67	33.67	n/a