Spatial Reasoning on RefSpatial-Bench
72.2Overall ScoreSeed
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| SeedModel Version/Scale=2.02026.04 | 72.2 | — | — | |
| QwenModel Version/Scale=3.5 A17B2026.04 | 61 | — | — | |
| HY-EmbodiedModel Version/Scale=0.5 MoE A32B2026.04 | 57.2 | — | — | |
| RoboBrain-32B-2.0number of parameters=32B, version=2.02026.01 | 54 | 54 | 54 | |
| Lumo-1-Stage1training_stage=Stage 12025.12 | 51 | 51.99 | 50 | |
| Lumo-1-Stage2training_stage=Stage 22025.12 | 50.75 | 49.5 | 52 | |
| KimiModel Version/Scale=K2.52026.04 | 43.3 | — | — | |
| Action-Sketcher-3Bnumber of parameters=3B2026.01 | 42.5 | 44.5 | 41 | |
| RoboBrain-7B-2.0parameters=7B2025.12 | 41.07 | 46.1 | 36.03 | |
| Gemini-2.5-Pro-preview-05-062026.01 | 38.16 | 44.58 | 31.73 | |
| GeminiModel Version/Scale=3.0 Pro2026.04 | 33.2 | — | — | |
| RoboBrain-7B-2.0number of parameters=7B, version=2.02026.01 | 32.5 | 36 | 29 | |
| Gemini-2.5-Flash-preview-04-172026.01 | 30.25 | 37.5 | 23 | |
| Qwen2.5-VL-72B-Instructnumber of parameters=72B2026.01 | 19.67 | 23.5 | 15.83 | |
| GPT-o4-mini-2025-05-162026.01 | 17.29 | 15 | 19.58 | |
| Qwen2.5-VL-32B-Instructnumber of parameters=32B2026.01 | 13.72 | 16.83 | 10.6 | |
| Qwen2.5-VL-32B-Instructparameters=32B, type=Instruct2025.12 | 11.24 | 12.99 | 9.5 | |
| RoboBrain-7B-1.0number of parameters=7B, version=1.02026.01 | 9.92 | 14.43 | 5.41 | |
| GPT-4o-2024-11-202026.01 | 8.78 | 8 | 9.55 | |
| Claude-Sonnet-4-2025-05-142026.01 | 7.69 | 5 | 10.37 | |
| Qwen2.5-VL-7B-Instructparameters=7B, type=Instruct2025.12 | 6.96 | 10.47 | 3.46 | |
| Cosmos-Reason1-7Bnumber of parameters=7B2026.01 | 5.44 | 9.84 | 1.04 | |
| Magma-8Bnumber of parameters=8B2026.01 | 4.5 | 1 | 8 | |
| VeBrain-8Bnumber of parameters=8B2026.01 | 0.3 | 0.03 | 0.57 |