LMGame-Bench Candy Crush — leaderboard
LLM game-playing benchmark: Candy Crush testing pattern matching and strategic swap selection.
Metric: Score. Source: huggingface.co. Status: saturation imminent. 25 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | O3 (2025-04-16) | 647 |
| 2 | Grok 4 | 616.7 |
| 3 | GPT-5 (High) | 615.7 |
| 4 | DeepSeek R1 0528 | 491.7 |
| 5 | O4 Mini (2025-04-16) | 487.3 |
| 6 | Claude 3.7 Sonnet (20250219) | 484 |
| 7 | Claude Sonnet 4 (20250514) | 478.3 |
| 8 | Claude Opus 4 (20250514) | 464 |
| 9 | DeepSeek R1 | 447.3 |
| 10 | GPT-OSS-120B | 441.3 |
| 11 | Gemini 2.5 Pro (Preview 05-06) | 416.3 |
| 12 | Qwen 3 235B A22B FP8 | 363.3 |
| 13 | Gemini 2.5 Flash (Preview 04-17) | 334.7 |
| 14 | GLM-4.5 | 324.3 |
| 15 | Grok 3 Mini Beta | 254 |
Interactive version: theaggregate.ai/benchmark?slug=lmgame-bench-candy-crush · How the rankings work · Data refreshed daily, snapshot 2026-07-22.