2048 Game on 2048 3072 target variant
6,962ScoreGPT-5
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-52026.02 | 6,962 | |
| Base Model + GLAD + MC-CriticBackbone=Qwen3-4B-Instruct2026.02 | 6,013.7 | |
| Base Model + GLADBackbone=Qwen3-4B-Instruct2026.02 | 4,565.7 | |
| Doubao-Seed-1.82026.02 | 4,210 | |
| UITARS-1.52026.02 | 3,920 | |
| Qwen3-235B-A22B-Instruct2026.02 | 2,688 | |
| Doubao-Seed-1.62026.02 | 2,054 | |
| Qwen3-30B-A3B-Instruct2026.02 | 1,740 | |
| Qwen3-4B-Instruct (Base Model)2026.02 | 1,603 | |
| Claude-4.5-Sonnet2026.02 | 120 |