Visual Tool-Use on MME-RealWorld
65.8AccuracyARM-Thinker-7B
Evaluation Results
| Method | Links | |
|---|---|---|
| ARM-Thinker-7BSize=7B, Backbone=Qwen2.5-VL-7B2025.12 | 65.8 | |
| Mini-o3Source=Lai et al. (2025)2025.12 | 65.5 | |
| Pixel ReasonerSource=Lai et al. (2025)2025.12 | 64.4 | |
| DeepEyesSource=Lai et al. (2025)2025.12 | 64 | |
| Qwen3-VL-8BSize=8B2025.12 | 63.1 | |
| InternVL3.5-8BSize=8B2025.12 | 62.8 | |
| InternVL3-8BSize=8B2025.12 | 61.2 | |
| Qwen2.5-VL-7BSize=7B2025.12 | 58.5 | |
| GPT-4oSource=Lai et al. (2025)2025.12 | 45.2 |