Needle-in-a-haystack on NIAH 1
79.69Success Rate (1k Context)Avey-B
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Avey-BScale=Large2026.02 | 79.69 | 79.24 | 79.03 | 79.58 | 79.44 | 78.44 | 76.76 | 76.06 | |
| NeoBERTScale=Medium2026.02 | 79.65 | 79.13 | 74.73 | — | — | — | — | — | |
| Avey-BScale=Base2026.02 | 79.41 | 79.21 | 78.94 | 79.19 | 78.91 | 77.73 | 77.18 | 75.72 | |
| ModernBERTScale=Large2026.02 | 68.8 | 67.52 | 67.2 | — | — | — | — | — | |
| ModernBERTScale=Base2026.02 | 67.74 | 67.64 | 68.31 | 70.67 | — | — | — | — |