Multi-modal Visualization Generation on MultiVis-Bench Iterative Refinement benchmark leaderboard · SOTA2 Research

Multi-modal Visualization Generation on MultiVis-Bench Iterative Refinement

66.67Low-Level Type ScoreLLM Workflow

Evaluation Results

Method	Links
LLM Workflow2026.01		66.67	68.84	38.41	71.01	78.26	83.33	67.75	89.86	88.33	78.8	80.25	72.28	89.86	81.62	74.69
MultiVis-Agent2026.01		63.04	49.28	36.96	70.29	78.26	76.81	62.44	87.86	87.54	81.52	73.19	72.64	92.39	80.6	71.52
LLM Workflow2026.01		61.59	60.14	25.36	66.67	60.87	73.91	58.09	81.7	83.77	75.54	69.57	65.76	86.52	75.25	66.67
Instructing LLM2026.01		60.87	68.12	33.33	68.12	71.01	78.26	63.29	78.8	76.96	72.64	71.92	66.49	82.54	73.85	68.57
Instructing LLM2026.01		57.25	63.04	25.36	65.22	63.04	73.19	57.85	75.18	78.55	72.46	68.3	63.22	82.54	71.43	64.64
MultiVis-Agent2026.01		55.8	25.36	19.57	65.94	65.94	68.12	50.12	79.89	79.86	72.83	54.17	62.14	88.26	70.18	60.15
nvAgent2026.01		19.57	6.52	0	51.45	57.97	50.72	31.04	25.36	25.22	24.64	13.41	21.74	47.83	24.12	27.58
nvAgent2026.01		14.49	11.59	0	34.78	36.96	33.33	21.86	23.55	22.32	20.83	15.4	18.66	33.19	20.95	21.4