leaderboard.md

model_name	score	VSI [66]	SITE [57]	MMSI [68]	OmniSpatial [23]	MindCube ∗ [69]	STARE [32]	CoreCognition [33]	SpatialViz [55]	source_title	source_url	notes
Human	79.2	79.2	67.5	97.2	92.63	94.55	96.5	86.98	82.46	EASI	https://arxiv.org/pdf/2508.13142.pdf	None
Qwen3-8B-Instruct [65]	57.9	57.9	45.83	31.1	45.73	29.42	39.76	69.67	17.54 †	EASI	https://arxiv.org/pdf/2508.13142.pdf	† indicates cases where generations were truncated due to overlong chains of thought, yielding no final answer; such instances are counted as incorrect, which depresses the score.
InternVL3.5-8B [56]	56.05	56.05	43.79	27.3	46.71	42.5	40.18	66.4	23.98	EASI	https://arxiv.org/pdf/2508.13142.pdf	None
GPT-5-2025-08-07 [45]	55.03	55.03	61.88	41.8	59.9	56.3	54.59	84.37	51.27	EASI	https://arxiv.org/pdf/2508.13142.pdf	None
Gemini-2.5-pro-2025-06 [52]	53.57	53.57	57.06	38	55.38	57.6	49.14	76.7	42.71	EASI	https://arxiv.org/pdf/2508.13142.pdf	None
Seed-1.6-2025-06-15 [51]	49.91	49.91	54.61	38.3	49.32	48.75	46.06	77.17	34.58	EASI	https://arxiv.org/pdf/2508.13142.pdf	None
GPT-5-mini-2025-08-07 [45]	48.67	48.67	52.47	34.1	55.52	56.69	52.51	77.77	44.66	EASI	https://arxiv.org/pdf/2508.13142.pdf	None
Grok-4-2025-07-09 [62]	47.92	47.92	47.01	37.8	46.84	63.56	26.9	79.27	19.40 †	EASI	https://arxiv.org/pdf/2508.13142.pdf	† indicates cases where generations were truncated due to overlong chains of thought, yielding no final answer; such instances are counted as incorrect, which depresses the score.
InternVL3-78B [79]	47.55	47.55	52.72	30.5	50.95	49.52	42	71.16	31.10	EASI	https://arxiv.org/pdf/2508.13142.pdf	None
GPT-5-nano-2025-08-07 [45]	43.22	43.22	35.81	28.9	47.81	41.48	46.05	67.92	35.59	EASI	https://arxiv.org/pdf/2508.13142.pdf	None
InternVL3-8B [79]	42.14	42.14	41.15	28	46.25	41.54	41.36	60.92	30.00	EASI	https://arxiv.org/pdf/2508.13142.pdf	None
Qwen2.5-VL-72B-Instruct [1]	35.77	35.77	47.41	32.5	47.81	42.4	38.37	69.22	32.54	EASI	https://arxiv.org/pdf/2508.13142.pdf	None
Random Choice	34	34	0	25	24.98	32.35	34.8	33.93	25.08	EASI	https://arxiv.org/pdf/2508.13142.pdf	VSI random choice here is chance level(Frequency).
Qwen2.5-VL-7B-Instruct [1]	32.3	32.3	37.64	26.8	39.07	36.05	35.03	62.16	26.78	EASI	https://arxiv.org/pdf/2508.13142.pdf	None
Qwen2.5-VL-3B-Instruct [1]	27	27	33.14	28.6	42.47	37.6	37.83	60.19	21.86	EASI	https://arxiv.org/pdf/2508.13142.pdf	None