Benchmarks
347 sources · 1,258 models × 3,038 benchmarks · 158,912 scoresINDUCTION v1 tests synthesis of one first-order concept formula from finite relational worlds. Its 775 solver-checkable instances span fully observed exact induction, contrastive induction, and existential completion under partial observation.
No signal or difficulty estimate. No usable scores.
This is a benchmark-family page. Scores are recorded on the component benchmarks below rather than directly on the parent, because the components use different score scales and directions.
| Metric | Models | Leader | Direction | Date |
|---|---|---|---|---|
| Concept Synth INDUCTION v1 - CI | 13 | GPT-5.282.5 | higher is better | 2026-05-23 |
| Concept Synth INDUCTION v1 - EC | 13 | GPT-5.494.0 | higher is better | 2026-05-23 |
| Concept Synth INDUCTION v1 - FullObs | 13 | Grok 450.7 | higher is better | 2026-05-23 |