Instruction Following
Benchmarks
DatasetSOTA methodMetricTrendResultsLast updated
73.92Language Democratization
4
Feb 27, 2026
58.76Win Rate (vs. Clust Rand)
4
Feb 26, 2026
59.13Win Rate (vs. Clust Rand)
4
Feb 26, 2026
0.29Drift (pp)
3
Jun 23, 2026
28.1Strict Score
3
Jun 18, 2026
64.4R Rate
3
Jun 8, 2026
61.2R Score
3
Jun 8, 2026
74.5Response Rate
3
Jun 8, 2026
56.93Strict Accuracy (Prompt-level)
3
Jun 4, 2026
83.8CSR
3
Jun 4, 2026
78.6CSR
3
Jun 4, 2026
88.7CSR (Correctness Score Rate)
3
Jun 4, 2026
38.17Accuracy
3
May 14, 2026
62.8SEIF Win Rate
3
May 11, 2026
9.11MT-Bench Score
3
May 7, 2026
93.2Win Rate
3
Feb 26, 2026
78.36Win Rate
3
Feb 26, 2026
669Score A
3
Feb 26, 2026
14Success Rate (Sponge)
2
Jul 7, 2026
0.794TRANS-COMET
2
Jul 3, 2026
0.765TRANS-COMET
2
Jul 3, 2026
0.763TRANS-COMET
2
Jul 3, 2026
53.1QA-BERTScore
2
Jul 3, 2026
0.02Cost (cents)
2
Jun 26, 2026
56.26Original Ratio
2
Jun 18, 2026