Multimodal Image Understanding on LLaVA Evaluation Suite 1.5 (test val) benchmark leaderboard · SOTA2 Research

Multimodal Image Understanding on LLaVA Evaluation Suite 1.5 (test val)

63.5GQABaseline

Evaluation Results

Method	Links
Baseline2026.07		63.5	68.21	62.63	1,787.33	88.52	72.38	48.91	56.14	35.7	100
TORINO-P2026.07		60.13	66.15	60.4	1,753.61	87.86	71.74	46.61	55.51	34.4	97.23
TORINO-M2026.07		59.92	66.32	60.31	1,728.88	87.94	71.64	46.53	55.56	34.4	97.04
FOLDER2026.07		59.8	65.81	59.97	1,678.18	88.38	71.59	42.79	55.46	34.6	95.8
PruMerge2026.07		59.22	65.98	61.34	1,729.59	87.04	72.24	43.07	55.63	35.8	96.69
PruneSID2026.07		59.15	65.98	60.22	1,740.8	88.06	71.49	43.3	55.36	34.8	96.26
Random2026.07		58.8	62.54	58.08	1,781.1	86.49	70.95	36.04	55.26	33.4	93.12
TORINO-P2026.07		58.62	65.46	59.97	1,745.8	86.96	70.8	43.93	55.97	34.5	95.99
PruneSID2026.07		58.43	63.75	58.93	1,731.43	86.86	72.24	41.23	55.78	32.3	94.28
TORINO-M2026.07		58.2	65.21	59.97	1,725.12	86.5	71.1	43.98	56.19	34.6	95.82
FOLDER2026.07		58.01	61.68	56.44	1,693.52	86.89	71.24	35.88	56.1	30.3	91.27
Random2026.07		57.29	58.93	54.47	1,774.34	84.5	70.6	30.06	55.67	27.8	88.26
PruMerge2026.07		57.25	66.07	59.36	1,676.44	84.14	72.43	42.49	55.86	33.9	94.67
PruneSID2026.07		57.03	60.4	55.67	1,727.21	85.98	68.57	39.88	58.29	30.2	91.74
TORINO-P2026.07		55.76	62.37	56.79	1,676.53	80.74	69.51	35.66	57.43	31.7	90.55
TORINO-M2026.07		55.48	61.51	55.84	1,666.57	80.44	69.41	35.76	57.72	31.5	90.1
FOLDER2026.07		54.68	54.21	46.82	1,642.95	82.48	67.72	26.94	58.19	26	83.4
Random2026.07		54.12	51.72	46.39	1,640.23	77.98	65.84	21.17	56.88	28.7	81.21
PruMerge2026.07		53.74	61.77	56.27	1,600.72	73.89	71.54	39.71	57.24	28.9	89