Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

Long-context understanding on InfiniteBench

0.6812En. MC Accuracy

FlexPrefill

0.3179280.4122390.506550.600861Feb 3, 2026
Updated 4d ago

Evaluation Results

MethodLinks
2026.02
0.68120.27370.110.5080.99150.96950.22570.2030.4953
2026.02
0.67250.2680.1850.5160.97460.9780.22290.19540.5016
2026.02
0.67250.2760.140.4680.98980.96270.22860.20050.4923
2026.02
0.65940.27610.1850.4960.97460.9780.21430.19290.497
2026.02
0.65070.28010.170.550.98470.97460.25710.2030.5088
2026.02
0.64190.27640.190.5520.98470.9780.240.20560.5086
2026.02
0.38430.1630.05500.33730.30.05140.26140.1941
2026.02
0.37990.16760.05500.34580.31190.060.26140.1977
2026.02
0.35810.17450.0700.41190.65760.04570.26650.248
2026.02
0.35370.16710.06500.39320.65930.03140.25630.2408
2026.02
0.33190.16870.05500.36610.62710.01430.27160.2293
2026.02
0.33190.16810.05500.35760.62880.00860.27410.228