Embodied Pointing and Visual Question Answering on Pointing and VQA Benchmarks Suite
71Where2Place ScoreSPARC (4B)
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| SPARC (4B)Parameter Scale=4B2026.06 | 71 | 79.1 | 54.5 | 44.9 | 61 | 86.4 | 35.3 | 65.7 | 62.7 | |
| RoboInter-LLaVAOV (7B)Parameter Scale=7B2026.06 | 66.3 | — | — | — | — | 89.3 | — | — | — | |
| RoboRefer (2B SFT)Parameter Scale=2B, Training Protocol=SFT2026.06 | 66 | — | 33.8 | — | 66.4 | 72.8 | — | 24.7 | — | |
| RoboInter-Qwen (7B)Parameter Scale=7B2026.06 | 65.8 | — | — | — | — | 85.6 | — | — | — | |
| MiMo-Embodied (7B)Parameter Scale=7B2026.06 | 63.6 | — | 48 | 46.7 | 61.8 | 82.3 | — | 46.9 | — | |
| RoboBrain 2.0 (7B)Parameter Scale=7B2026.06 | 63.5 | — | 54 | 30.3 | 54.2 | 70.4 | — | 26.67 | — | |
| SPARC (0.8B)Parameter Scale=0.8B2026.06 | 62 | 75.4 | 33.1 | 36.9 | 45.5 | 81 | 34.6 | 56.3 | 53.1 | |
| RoboInter-Qwen (3B)Parameter Scale=3B2026.06 | 58.3 | — | — | — | — | 80 | — | — | — | |
| SPARC (0.8B, VT-FT)Parameter Scale=0.8B, Training Protocol=VT-FT2026.06 | 58 | 76.7 | 37.1 | 40.4 | 53.5 | 81 | 30.3 | 48.3 | 53.2 | |
| Gemini 2.5 Pro2026.06 | 53 | — | 29.2 | 48.3 | — | 49.5 | — | 21.7 | — | |
| Qwen3.5-4BParameter Scale=4B, Training Protocol=Zero-shot2026.06 | 47 | 49.7 | 30.5 | 30.3 | 48.4 | 70.1 | 30.5 | 19 | 42.6 | |
| RoboPoint (13B)Parameter Scale=13B2026.06 | 46.8 | — | 16.1 | — | — | 49.8 | — | 19.1 | — | |
| FSD (13B)Parameter Scale=13B2026.06 | 45.8 | — | — | — | — | 56.7 | — | 61.8 | — | |
| Qwen2.5-VL (72B)Parameter Scale=72B2026.06 | 37.2 | — | 20.9 | — | — | 78.5 | — | 23.3 | — | |
| EO-1 (3B)Parameter Scale=3B2026.06 | — | — | — | 45.5 | — | — | 36.4 | — | — | |
| RynnBrain (2B)Parameter Scale=2B2026.06 | — | — | 52.7 | 42.3 | 65.7 | — | — | — | — | |
| RynnBrain (8B)Parameter Scale=8B2026.06 | — | — | 59.2 | 46.8 | 73.1 | — | — | — | — |