Multi-task long-context understanding on LongBench benchmark leaderboard · SOTA2 Research

Multi-task long-context understanding on LongBench

37.66QasperSparseD

Evaluation Results

Method	Links
SparseD2026.02		37.66	40.96	41.39	41.61	24.17	23.51	19.66	72.5	86.85	15.75	36.83	63.86	61.98	43.59
Fast-dLLM2026.02		37.54	43.24	35.56	35.74	17.97	21.14	19.57	70.5	88.25	16.75	46.17	62.21	61.11	42.75
Sparse-dLLM2026.02		37.5	43.23	36.83	34.97	17.05	20.6	20.05	70	88.38	17	46.5	62.8	61.2	42.78
Focus-dLLM2026.02		37.38	41.96	38.96	38.56	18.05	21.06	19.26	70	88.76	17.25	44.25	60.62	60.5	42.82
Vanilla2026.02		35.58	40.49	41.59	42.1	23.36	23.51	19.66	71.5	87.34	15.75	32.5	63.79	62.23	43.03
Vanilla2026.02		19.14	25.87	16.27	18	12.08	32.83	22.48	80	91.58	41	96.75	68.23	59.5	44.9
SparseD2026.02		19.09	25.87	15.45	18.04	11.92	32.64	22.5	79.5	90.7	41.5	96.79	68.1	59.02	44.7
Fast-dLLM2026.02		18.34	29.9	17.03	17.11	13.36	30.05	22.89	79.5	91.03	42	94.75	67.5	58.1	44.74
Sparse-dLLM2026.02		18.04	27.26	20.59	17.88	13.67	29.95	23.57	76.5	91.93	41.5	97.12	67.5	57.72	44.86
Focus-dLLM2026.02		17.02	29.11	22.47	21.49	20.2	26.75	21.45	77	90.78	41	95.73	66.72	57.14	45.14