Spatial Reasoning on BLINK-S
90.21AccuracyQwen2.5-VL-7B
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen2.5-VL-7Bparameters=7B, zero-shot=true2025.12 | 90.21 | |
| GPT-4o-minizero-shot=true2025.12 | 88.11 | |
| Qwen3-VL-8Bparameters=8B, zero-shot=true2025.12 | 86.71 | |
| InternVL-3.5-8Bparameters=8B, zero-shot=true2025.12 | 86.01 | |
| Cosmos-Reason1-7Bparameters=7B, zero-shot=true2025.12 | 84.62 | |
| Gemini-2.5-prozero-shot=true2025.12 | 84.62 | |
| Qwen2.5-VL-3Bparameters=3B, zero-shot=true2025.12 | 81.12 | |
| Claude-Sonnet-4zero-shot=true2025.12 | 79.72 | |
| RoboBrain2.0-7Bparameters=7B, zero-shot=true2025.12 | 79.02 | |
| GenieReasonerparameters=3B, zero-shot=true2025.12 | 74.83 |