Embodied vision-language reasoning on A2Eval B~
65.97ScoreQwen3-VL-235B-A22B-Thinking
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Qwen3-VL-235B-A22B-ThinkingParameters=235B-A22B, Variant=Thinking2026.02 | 65.97 | — | — | |
| Internvl-3.5-241B-A28BParameters=241B-A28B2026.02 | 65.68 | — | — | |
| GPT-5-20250807-MiniVersion=20250807, Size=Mini2026.02 | 65.52 | — | — | |
| Qwen3-VL-30B-A3B-ThinkingParameters=30B-A3B, Variant=Thinking2026.02 | 62.58 | — | — | |
| Internvl-3.5-38BParameters=38B2026.02 | 62.37 | — | — | |
| Qwen3-VL-235B-A22B-InstructParameters=235B-A22B, Variant=Instruct2026.02 | 61.88 | — | — | |
| Internvl-3.5-30B-A3BParameters=30B-A3B2026.02 | 61.79 | — | — | |
| Qwen3-VL-30B-A3B-InstructParameters=30B-A3B, Variant=Instruct2026.02 | 61.47 | — | — | |
| Internvl-3.5-8BParameters=8B2026.02 | 55.98 | — | — | |
| Qwen2.5-VL-72B-InstructParameters=72B, Variant=Instruct2026.02 | 54.76 | — | — | |
| Qwen2.5-VL-32B-InstructParameters=32B, Variant=Instruct2026.02 | 50.77 | — | — | |
| Qwen2.5-VL-7B-InstructParameters=7B, Variant=Instruct2026.02 | 48.22 | — | — | |
| Qwen2.5-VL-3B-InstructParameters=3B, Variant=Instruct2026.02 | 39.39 | — | — | |
| Ranking correlation between Our (B~) and Source (B)Type=Correlation Analysis2026.02 | — | 0.94 | 0.81 | |
| Ranking correlation with Human (H)Type=Correlation Analysis2026.02 | — | 0.85 | 0.72 |