Report-based data analysis on Deep Data Research (test) benchmark leaderboard · SOTA2 Research

Report-based data analysis on Deep Data Research (test)

39.15Accuracy (Sample Avg, Msg, MIMIC)GPT-5-2

Evaluation Results

Method	Links
GPT-5-22026.06		39.15	62.21	65.97	39.77	70.56	55.94	37.2	62.15	65.93	37.72	70.46	55.57	55.22
GPT-5-22026.06		37.59	53.85	63.15	35.28	70.64	48.8	35.47	53.54	63.73	33.56	70.46	48.35	51.2
GPT-5-22026.06		37.2	49.93	60.16	37.07	59.63	48.52	35.12	49.85	59.65	34.78	59.69	48.51	48.34
Claude 4.5 Sonnet2026.06		36.75	58.68	76.55	36.99	72.92	65.63	34.6	58.77	76.45	34.08	72.92	65.62	57.5
Claude 4.6 Sonnet2026.06		36.18	60.64	77	37.05	58.92	66.34	34.26	60.62	76.92	35.12	59.08	66.41	55.71
Claude 4.5 Sonnet2026.06		36.09	58.46	70.5	40.54	62.13	57.58	34.08	58.15	70.96	38.58	61.85	57.93	53.9
Qwen3.5-397B-A17B2026.06		35.78	64.02	68.22	41.5	73.95	65.56	33.56	63.69	68.45	39.27	73.85	66.25	57.84
Claude 4.5 Sonnet2026.06		34.9	56.54	75.32	34.92	56.18	54.01	32.53	56.62	75.35	33.04	56.31	54.47	51.68
Qwen3.5-397B-A17B2026.06		32.63	53.33	58.57	37.45	58.5	40.68	30.97	53.23	58.56	35.12	58.15	40.97	46.51
DeepSeek-V4-Pro2026.06		32.18	57.79	76.69	43.77	66.42	71.54	30.62	57.85	76.3	42.39	66.46	71.9	57.83
DeepSeek-V4-Pro2026.06		30.51	55.76	67.51	35.61	60.88	57.72	28.72	56	67.66	33.56	60.92	58.08	51.08
Qwen3.5-397B-A17B2026.06		30.49	49.78	50.02	30.04	45.07	27.99	28.2	49.54	49.29	28.2	44.62	28.1	38.45
DeepSeek-V4-Pro2026.06		28.32	59	69.84	43.27	65.81	58.46	25.78	59.38	69.86	40.66	65.85	58.71	53.75