LMGame-Bench Super Mario Bros — leaderboard
LLM game-playing benchmark: Super Mario Bros level completion testing spatial reasoning and real-time decision making.
Metric: Score. Source: huggingface.co. 12 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | O3 (2025-04-16) | 3445 |
| 2 | GPT-4.1 | 2126.3 |
| 3 | GPT-4o (2024-11-20) | 2047.3 |
| 4 | Gemini 2.5 Pro (Preview 05-06) | 1498.3 |
| 5 | Llama 4 Maverick Instruct FP8 | 1468.7 |
| 6 | O4 Mini (2025-04-16) | 1448 |
| 7 | GPT-5 (High) | 1432.3 |
| 8 | Claude 3.7 Sonnet (20250219) | 1418.7 |
| 9 | Gemini 2.5 Flash (Preview 04-17) | 1385 |
| 10 | Claude 3.5 Sonnet (20241022) | 1267.7 |
| 11 | O1 (2024-12-17) | 855 |
Interactive version: theaggregate.ai/benchmark?slug=lmgame-bench-super-mario-bros · How the rankings work · Data refreshed daily, snapshot 2026-07-22.