Novelty Bench
Allen Institute for AI publishes Novelty Bench GPT-5.1 is used in Novelty Bench evaluation GPT-5.2 is used in Novelty Bench evaluation GPT-5.4 is used in Novelty Bench evaluation Claude Sonnet 4.5 is used in Novelty Bench evaluation Claude Opus 4.5 is used in Novelty Bench evaluation Claude Opus 4.6 is used in Novelty Bench evaluation ICLR 2026 provided review data for Novelty Bench Novelty Bench is posted on Hugging Face· Entity summary generated by Graphiti
Relation timeline
BY VALID-FROM DATE- Novelty BenchPOSTED ONHugging Face
Novelty Bench is posted on Hugging Face
- GPT-5.1USED INNovelty Bench
GPT-5.1 is used in Novelty Bench evaluation
- GPT-5.2USED INNovelty Bench
GPT-5.2 is used in Novelty Bench evaluation
- Claude Opus 4.6USED INNovelty Bench
Claude Opus 4.6 is used in Novelty Bench evaluation
- Allen Institute for AIPUBLISHESNovelty Bench
Allen Institute for AI publishes Novelty Bench
- GPT‑5.4USED INNovelty Bench
GPT-5.4 is used in Novelty Bench evaluation
- ForethoughtPROVIDED REVIEW DATA FORNovelty Bench
ICLR 2026 provided review data for Novelty Bench
- My obituary for programmingUSED INNovelty Bench
Claude Sonnet 4.5 is used in Novelty Bench evaluation
- Claude Opus 4.5USED INNovelty Bench
Claude Opus 4.5 is used in Novelty Bench evaluation
Relations are facts Graphiti extracted from reports, not verbatim quotes. An end date means later reporting showed the relation no longer holds.