Long-context downstream tasks on LongBench-E (test) benchmark leaderboard · SOTA2 Research

Long-context downstream tasks on LongBench-E (test)

49.92S-Doc QA PerfElastic Attention

Evaluation Results

Method	Links
Elastic Attention2026.01		49.92	0.64	48.92	0.63	30.14	0.73	67.99	0.74	54	0.64	60.7	0.79	53.35	0.69
Elastic Attention2026.01		49.4	0.71	52.94	0.69	30.3	0.8	68.55	0.79	49.66	0.72	56.49	0.87	52.71	0.77
Llama-3.1-8B-Instruct2026.01		48.75	—	51.85	—	30.26	—	68.16	—	56	—	55.81	—	53.28	—
DuoAttention2026.01		48.65	0.7	45.21	0.7	29.93	0.7	67.35	0.7	54.56	0.7	56.47	0.7	51.82	0.7
PruLong2026.01		47.69	0.7	43.05	0.7	30.02	0.7	67.86	0.7	53.56	0.7	61.07	0.7	52.11	0.7
Elastic Attention2026.01		46.15	0.64	46.54	0.65	28.19	0.72	67.52	0.71	48.07	0.65	62.95	0.78	51.51	0.69
PruLong2026.01		46.05	0.7	46.88	0.7	28.3	0.7	66.61	0.7	48.66	0.7	62.24	0.7	51.34	0.7
Qwen3-8B2026.01		45.57	—	51.59	—	28.34	—	66.64	—	50.16	—	61.2	—	52.16	—
DuoAttention2026.01		45.45	0.7	44.52	0.7	28.16	0.7	66.42	0.7	47.5	0.7	62.38	0.7	50.67	0.7
Elastic Attention2026.01		44.4	0.68	39.42	0.71	28.49	0.82	65.26	0.76	44.35	0.74	54.29	0.87	47.59	0.76
Elastic Attention2026.01		44.01	0.75	49.99	0.76	28.3	0.83	66.23	0.8	50.95	0.77	60.57	0.86	51.66	0.8
InfLLM-V22026.01		43.77	—	46.3	—	30.08	—	67.32	—	42.03	—	64.3	—	50.73	—
Qwen3-4B2026.01		43.69	—	38.48	—	28.46	—	66.21	—	49.59	—	54.38	—	48.45	—
InfLLM-V22026.01		43.3	—	34.97	—	29.95	—	64.19	—	38.54	—	59.58	—	46.68	—
Elastic Attention2026.01		42.2	0.66	38.86	0.68	28.5	0.76	65.73	0.73	48.43	0.71	54.34	0.82	48.08	0.73
InfLLM-V22026.01		42.2	—	42.33	—	29.58	—	64.55	—	45.87	—	59.58	—	49.03	—
DuoAttention2026.01		41.73	0.7	35.72	0.7	28.47	0.7	64.59	0.7	47.17	0.7	53.91	0.7	46.95	0.7
PruLong2026.01		41.58	0.7	37.58	0.7	28.53	0.7	64.8	0.7	47.23	0.7	53.2	0.7	47.19	0.7