Claude Sonnet 4.5 (High): benchmark results
Claude Sonnet 4.5 evaluated at the high reasoning-effort setting. Provider: Anthropic. Released 2025-09-29. Access: API.
Aggregate Rating 1710 ± 20, rank #327 of 2168 rated models, from 96 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Clembench Text v3.0 | 90.1 | clemscore | 100 |
| HAL GAIA Level 2 | 74.42 | Accuracy (%) | 100 |
| HAL SWE-bench Verified Mini | 72 | Score (%) | 100 |
| Horangi 4 - Ko-HellaSwag | 90 | Accuracy (%) | 99 |
| Horangi 4 - KorQuAD 1.0 | 90.05 | F1 (x100) | 99 |
| Horangi 4 - GLP - Logical Reasoning | 89.5 | Score (%) | 97.1 |
| Horangi 4 - Ko-TruthfulQA | 96 | Accuracy (%) | 96.6 |
| Horangi 4 - Ko-MT-Bench | 89.38 | Judge rating (1-10, x10) | 94.2 |
| HAL GAIA | 70.91 | Accuracy (%) | 93.8 |
| HAL GAIA Level 1 | 77.36 | Accuracy (%) | 93.8 |
| Horangi 4 - Ko-Moral | 84 | Accuracy (%) | 93.8 |
| IKP - T2 | 100 | Accuracy (%) | 92.6 |
Interactive version: theaggregate.ai/model?slug=claude-sonnet-4-5-high · How It Works · Data refreshed daily, snapshot 2026-10-20.