Claude 4.5: benchmark results
Provider: Anthropic. Access: API.
Aggregate Rating 1687 ± 39, rank #409 of 10916 rated models, from 13 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| LogicScore - MuSiQue Completeness | 63.88 | Completeness (%; answers whose extracted propositions link t | 100 |
| LogicScore - 2WikiMultiHopQA Completeness | 85.86 | Completeness (%; answers whose extracted propositions link t | 95 |
| LogicScore - HotpotQA Completeness | 78.33 | Completeness (%; answers whose extracted propositions link t | 95 |
| StatABench | 66.3 | Total (self-reported) | 87.5 |
| Executable Games | 73.63 | Success Rate (%; clean games, mean over five difficulty leve | 83.3 |
| LogicScore - HotpotQA Determinateness | 98.07 | Determinateness (%; answers whose own short answer o4-mini r | 75 |
| LogicScore - MuSiQue Determinateness | 89.66 | Determinateness (%; answers whose own short answer o4-mini r | 75 |
| LogicScore - 2WikiMultiHopQA Determinateness | 92.03 | Determinateness (%; answers whose own short answer o4-mini r | 70 |
| QUIET | 8.03 | QUIET Total (self-reported) | 40.9 |
| LogicScore - MuSiQue Essentiality | 27.13 | Essentiality (%; share of the answer's extracted proposition | 40 |
| LogicScore - 2WikiMultiHopQA Essentiality | 40.13 | Essentiality (%; share of the answer's extracted proposition | 37.5 |
| LogicScore - HotpotQA Essentiality | 36.4 | Essentiality (%; share of the answer's extracted proposition | 25 |
Interactive version: theaggregate.ai/model?slug=claude-4-5 · How It Works · Data refreshed daily, snapshot 2026-11-08.