SOTA Video Spatial Reasoning on Video-RoboSpatial and PapersWithCode

92.1Configuration Score

Human Level

Updated 2mo ago

Evaluation Results

Method	Links
Human Level 2026.03		92.1
Thinking with Spatial Code 2026.03		67
Thinking with Spatial Code + 2D box 2026.03		67
Thinking with Spatial Code (w/o RL) 2026.03		65.3
Thinking with Spatial Code (w/o RL) + 2D box 2026.03		65.3
GPT-5 2026.03		60.3
LLaVA-Video 2026.03		58
SpaceR 2026.03		56
LLaVA-OneVision 2026.03		56
Seed-1.6 2026.03		54.3
LLaVA-OneVision 2026.03		54.3
Gemini-2.5-Pro 2026.03		53.3
GPT-4o 2026.03		53
Qwen3-VL 2026.03		52.7
Qwen3-VL + 2D box 2026.03		52.7
LLaVA-Video 2026.03		52
Qwen2.5-VL 2026.03		49.7
SpatialLadder 2026.03		49.3
Spatial-MLLM 2026.03		49
Qwen3-VL 2026.03		48