List your company

Long-Form Question Answering on LFQA (GPT-4o Evaluation)

4.115GPT-4o ScoreNo watermark
3.913243.965624.0184.07038Apr 15, 2026

Evaluation results

MethodLinks
No watermarkMaximum generation length (T)=300 tokens, Minimum output length filter=50 tokens, Number of sampled prompts=5002026.044.115
QuantileMarkMaximum generation length (T)=300 tokens, Minimum output length filter=50 tokens, Number of sampled prompts=5002026.044.109
StealthInkMaximum generation length (T)=300 tokens, Minimum output length filter=50 tokens, Number of sampled prompts=5002026.044.047
MPACMaximum generation length (T)=300 tokens, Minimum output length filter=50 tokens, Number of sampled prompts=5002026.043.921