Visual Reasoning on SEED-Bench 2-Plus
72AccuracyGPT-4o
Evaluation results
| Method | Links | |
|---|---|---|
| GPT-4oReasoning Paradigm=Zero-Shot VLMs2026.01 | 72 | |
| VL-RethinkerReasoning Paradigm=Tool-use & RL Enhanced Reasoning2026.01 | 70.27 | |
| LaserReasoning Paradigm=Latent Reasoning2026.01 | 70.05 | |
| InternVL3.5-8BReasoning Paradigm=Zero-Shot VLMs2026.01 | 69.78 | |
| DeepEyesReasoning Paradigm=Tool-use & RL Enhanced Reasoning2026.01 | 69.08 | |
| Vision-R1Reasoning Paradigm=Tool-use & RL Enhanced Reasoning2026.01 | 68.95 | |
| MonetReasoning Paradigm=Latent Reasoning2026.01 | 65.88 | |
| Qwen2.5-VL-7BReasoning Paradigm=Zero-Shot VLMs2026.01 | 65.31 | |
| LLaVA-OneVisionReasoning Paradigm=Zero-Shot VLMs2026.01 | 61.22 | |
| PAPOReasoning Paradigm=Tool-use & RL Enhanced Reasoning2026.01 | 54.11 | |
| LVRReasoning Paradigm=Latent Reasoning2026.01 | 47.39 |