Multi-frame Spatial Understanding on MultiSPA benchmark leaderboard · SOTA2 Research

Multi-frame Spatial Understanding on MultiSPA

56.11Average ScoreMulti-SpatialMLLM

Evaluation Results

Method	Links
Multi-SpatialMLLM2025.05		56.11	74	75.33	49	90	90.83	45.5	85.89	42.33	18	40.42	12.92	49.11
Gemini-2.02025.05		30.31	57	28.67	5.67	73	62.17	16.34	51.89	14	0.33	9.42	2.33	42.89
GPT-4o2025.05		28.87	54.84	22.5	2	67.67	58.84	17.5	54.78	13.67	0	8.92	5.25	40.44
Claude-3.52025.05		27.5	38.17	34.84	1.33	54.67	62.17	10.5	55.11	16.33	0.33	8.5	1.92	46.11
InternVL-26B2025.05		21.36	50.5	2.5	1.67	44	49.34	5.17	50.22	13	0.67	8.75	3.58	26.89
InternVL-8B2025.05		20.43	49.5	3.34	1.67	33.33	48.17	3.34	52.33	14.33	0.33	8.84	2.5	27.45
SpatialRGPT2025.05		16.24	52.17	3	0	30.67	54.67	4.84	53.22	7.33	0	2.57	0	22.33