List your company

Idea Novelty Evaluation on 100 ICLR 2026 problem seeds (test)

3.73Mean NoveltyGPT-5.5
2.26362.64433.0253.4057Jul 5, 2026

Evaluation results

MethodLinks
GPT-5.5configuration=bare2026.073.730.5322772145
IdeaSparkbackbone=Opus-4.82026.072.920.53153216300
Opus-self-genbackbone=Opus-4.8, retrieval=live retrieval2026.072.860.5159222180
Opus-4.8configuration=bare, backbone=Opus-4.82026.072.320.61916811030