Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Long-context performance evaluation on RULER

96Accuracy

FP16

9.6832.0954.576.91Feb 7, 2026Mar 3, 2026Mar 27, 2026Apr 20, 2026May 14, 2026Jun 7, 2026Jul 1, 2026
Updated 23d ago

Evaluation Results

MethodLinks
2026.07
96------
2026.07
96------
2026.02
95.31------
2026.02
95.31------
2026.02
95------
2026.02
94.06------
2026.07
94------
2026.07
94------
2026.02
93.64------
2026.02
93.13------
2026.02
89.69------
2026.02
87.5------
2026.02
82.81------
2026.02
81.5------
2026.07
72------
2026.07
20------
2026.07
13------
2026.02
-91.44-----
2026.02
-90.15-----
2025.02
-8690.19583.485.576.3
2025.02
-18.61521.524.416.915
2025.02
--27.1----
2025.02
-73.675.676.872.97567.7
2025.02
--66.5----
2025.02
-69.669.768.270.469.869.8
2025.02
-75.677.877.377.277.468.2