Long-context evaluation on LongBench (aggregate)
58.7EM
Full KV
Updated 18d ago
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
2026.07 | 58.7 | 63.4 | 32.8 | 48.6 | 38.2 | 2.91 | 24 | 1 | |
2026.07 | 58.3 | 63 | 32.5 | 48.1 | 70.4 | 2.06 | 6.2 | 3.9 | |
2026.07 | 57.1 | 61.8 | 31.7 | 46.4 | 60.9 | 2.29 | 6.9 | 3.5 | |
2026.07 | 56.8 | 61.5 | 31.6 | 46 | 59.7 | 2.31 | 7.2 | 3.3 | |
2026.07 | 56.6 | 61 | 31.3 | 45.6 | 65.5 | 2.18 | 6.6 | 3.6 | |
2026.07 | 56.4 | 61.2 | 31.4 | 45.8 | 63 | 2.24 | 7.4 | 3.2 | |
2026.07 | 55.9 | 60.7 | 31.1 | 45.2 | 61.8 | 2.28 | 7.9 | 3 | |
2026.07 | 54.4 | 59.1 | 30.5 | 44 | 58.4 | 2.35 | 8.7 | 2.8 | |
2026.07 | 53.8 | 58.6 | 30.2 | 43.7 | 57.1 | 2.39 | 9.1 | 2.6 | |
2026.07 | 51.3 | 57 | 29.4 | 42.1 | 55.6 | 2.43 | 9.8 | 2.4 |