Visual Planning on FrozenLake (Levels 3-8)
100Success Rate (Level 3)Gemini-3-Flash
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Gemini-3-FlashInput Modality=Visual, Prompting Strategy=Prompt-only, Task-specific training status=None2026.06 | 100 | 100 | 99 | 96 | 84 | 97 | 96 | |
| Gemini-2.5-ProInput Modality=Visual, Prompting Strategy=Prompt-only, Task-specific training status=None2026.06 | 99 | 97 | 81 | 46 | 44 | 48 | 69.2 | |
| Qwen3-VL-32B-ThinkingInput Modality=Visual, Prompting Strategy=Prompt-only, Task-specific training status=None2026.06 | 98 | 79 | 59 | 35 | 28 | 21 | 53.3 | |
| Kimi-K2.5Input Modality=Visual, Prompting Strategy=Prompt-only, Task-specific training status=None2026.06 | 93 | 77 | 75 | 54 | 48 | 51 | 66.3 | |
| GPT-5Input Modality=Visual, Prompting Strategy=Prompt-only, Task-specific training status=None2026.06 | 92 | 65 | 53 | 30 | 38 | 32 | 51.7 | |
| Gemini-2.5-FlashInput Modality=Visual, Prompting Strategy=Prompt-only, Task-specific training status=None2026.06 | 92 | 91 | 74 | 58 | 37 | 33 | 64.2 | |
| Symbolic-Input Qwen3-VL-8BInput Modality=Symbolic, Reason=Oracle Upper Bound2026.06 | 82 | 72 | 65 | 39 | 43 | 46 | 57.8 | |
| Symbolic-Input Qwen3-VL-4BInput Modality=Symbolic, Reason=Oracle Upper Bound2026.06 | 79 | 62 | 51 | 40 | 28 | 26 | 47.7 | |
| MGSD-8BInput Modality=Visual, Task-specific training status=MGSD Trained2026.06 | 78 | 61 | 58 | 39 | 35 | 40 | 51.8 | |
| Qwen3-VL-235B-A22B-InstructInput Modality=Visual, Prompting Strategy=Prompt-only, Task-specific training status=None2026.06 | 77 | 56 | 44 | 20 | 20 | 18 | 39.2 | |
| MGSD-4BInput Modality=Visual, Task-specific training status=MGSD Trained2026.06 | 76 | 56 | 60 | 27 | 27 | 24 | 45 | |
| GPT-4oInput Modality=Visual, Prompting Strategy=Prompt-only, Task-specific training status=None2026.06 | 61 | 38 | 28 | 15 | 12 | 13 | 27.8 | |
| Claude-4.5-HaikuInput Modality=Visual, Prompting Strategy=Prompt-only, Task-specific training status=None2026.06 | 49 | 41 | 29 | 17 | 10 | 6 | 25.3 | |
| Qwen3-VL-8B-InstructInput Modality=Visual, Prompting Strategy=Prompt-only, Task-specific training status=None2026.06 | 42 | 30 | 27 | 18 | 16 | 13 | 24.3 | |
| Qwen3-VL-4B-InstructInput Modality=Visual, Prompting Strategy=Prompt-only, Task-specific training status=None2026.06 | 33 | 29 | 19 | 5 | 3 | 5 | 15.7 | |
| InternVL3-8BInput Modality=Visual, Prompting Strategy=Prompt-only, Task-specific training status=None2026.06 | 30 | 18 | 14 | 10 | 7 | 9 | 14.7 | |
| Qwen2.5-VL-7B-InstructInput Modality=Visual, Prompting Strategy=Prompt-only, Task-specific training status=None2026.06 | 18 | 18 | 11 | 8 | 7 | 6 | 11.3 | |
| Qwen2.5-VL-3B-InstructInput Modality=Visual, Prompting Strategy=Prompt-only, Task-specific training status=None2026.06 | 14 | 8 | 11 | 7 | 4 | 4 | 8 | |
| LLaVA-OneVision-7BInput Modality=Visual, Prompting Strategy=Prompt-only, Task-specific training status=None2026.06 | 13 | 8 | 4 | 3 | 3 | 1 | 5.3 |