Long-form generation on WritingBench benchmark leaderboard · SOTA2 Research

Long-form generation on WritingBench

5.1ScoreLarge

Evaluation Results

Method	Links
Large2026.03		5.1	9.1	1
AgentCollab2026.03		5	9.59	2.43
RouteLLM2026.03		4.7	10.27	2.21
FrugalGPT2026.03		4.5	12.22	1.94
Small2026.03		4.4	17.25	3.2
Random2026.03		4.2	13.1	1.29