Share your thoughts, 1 month free Claude Pro on usSee more

Intuitive Physics Understanding on LikePhys 1.0 (test)

100PPE: Ball Collision

GEOPHYS

Updated 1mo ago

Evaluation Results

Method
GEOPHYS 2026.06	100	45	100	89	100	92	-	93	-	61	-	60	78.6	81	60	62
GEOPHYS 2026.06	100	38	100	87	100	92	-	97	-	84	-	43	80.8	81	83	65
GEOPHYS Ensemble (Majority) 2026.06	100	77	100	96	100	95	-	99	-	86	-	78	90.9	96	93	83
GEOPHYS Ensemble (OR) 2026.06	100	88	100	99	100	100	-	100	-	100	-	92	98.3	100	97	97
GEOPHYS 2026.06	92	68	72	81	97	75	-	92	-	66	-	57	78.3	76	93	72
GEOPHYS 2026.06	90	82	100	90	100	88	-	19	-	59	-	62	77.6	61	94	87
Lin. probe 2026.06	85	38	85	61	85	75	-	75	-	45	-	45	62.4	65	45	45
GEOPHYS 2026.06	80	83	100	86	100	78	-	89	-	55	-	65	78.2	94	51	57
Hunyuan 2026.06	67	52	75	78	50	64	-	62	-	32	-	56	56.4	23	60	62
AnimateDiff 2025.10	63.3	60	65	51.7	61.1	52.9	78.6	62	63.3	44	71.7	56	60.8	-	-	-
AnimateDiff SDXL 2025.10	63.3	66.7	38.3	70	68.9	47.1	82.9	54	60	46	43.3	32	56	-	-	-
LTX Vid. 2026.06	63	42	65	32	81	53	-	68	-	60	-	52	55.3	51	72	25
CogVideoX1.5–5B 2025.10	61.7	50	51.7	31.7	25.6	52.9	22.9	54	21.7	68	23.3	62	43.8	-	-	-
Mochi 2026.06	60	50	68	35	16	71	-	62	-	56	-	60	48.1	17	40	42
CogVidX-2B 2026.06	60	62	40	37	17	64	-	64	-	60	-	68	51.8	19	50	82
Wan2.1-14B 2026.06	58	43	38	87	57	37	-	44	-	36	-	40	56.2	66	85	83
CogVidX-5B 2026.06	57	58	38	35	17	61	-	64	-	60	-	70	50.2	19	47	77
Wan2.1-1B 2026.06	57	47	33	80	60	34	-	40	-	22	-	40	52	67	67	77
ZeroScope 2025.10	53.3	55	46.7	58.3	73.3	38.6	84.3	46	61.7	40	40	42	53.3	-	-	-
ModelScope 2025.10	51.7	53.3	46.7	66.7	78.9	40	84.3	40	61.7	40	33.3	38	52.9	-	-	-
ModelScope 2026.06	48	47	53	33	21	60	-	60	-	60	-	62	47.1	16	38	67
ZeroScope 2026.06	47	45	53	42	27	61	-	54	-	60	-	58	46.7	16	38	60
CogVideoX–5B 2025.10	43.3	41.7	61.7	65	83.3	38.6	81.4	36	53.3	40	23.3	30	49.8	-	-	-
Wan2.1-T2V-1.3B 2025.10	43.3	53.3	66.7	20	40	65.7	32.9	60	33.3	78	23.3	60	48	-	-	-
Wan2.1-T2V-14B 2025.10	41.7	56.7	61.7	13.3	43.3	62.9	34.3	56	15	64	16.7	60	43.8	-	-	-
Mochi 2025.10	40	50	31.7	65	84.4	28.6	82.9	38	60	44	58.3	40	51.9	-	-	-
CogVideoX-2B 2025.10	40	38.3	60	63.3	83.3	35.7	81.4	36	50	40	18.3	32	48.2	-	-	-
CogVidX1.5 2026.06	38	50	48	68	74	47	-	46	-	32	-	38	56.2	77	78	77
AnimDiff 2026.06	37	40	35	48	39	47	-	38	-	56	-	44	39.2	21	37	28
AnimSDXL 2026.06	37	33	62	30	31	53	-	46	-	54	-	68	44	17	40	57
LTX v0.9.5 2025.10	36.7	58.3	35	68.3	18.9	47.1	48.6	32	28.3	40	75	48	44.7	-	-	-
Hunyuan T2V 2025.10	33.3	51.7	25	21.7	50	35.7	77.1	38	40	68	38.3	44	43.6	-	-	-