Claude Sonnet 4 (20250514) (Thinking): benchmark results
Claude Sonnet 4 (20250514) evaluated with thinking enabled. Provider: Anthropic. Released 2025-05-22. Access: API.
Aggregate Rating 1639 ± 18, rank #567 of 2168 rated models, from 48 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| HELM Safety - Safety Scenarios | 98.07 | Mean score | 95.9 |
| HELM Safety - Harm Bench | 98 | LM Evaluated Safety score | 94.2 |
| UGI - Writing | 62.31 | Writing Score | 92.6 |
| Web-Bench | 24.3 | Pass@1 (%) | 89.4 |
| UGI - Natural Intelligence | 47.43 | NatInt Score | 87.5 |
| HELM Capabilities - GPQA | 70.63 | COT correct | 86.6 |
| HELM Capabilities - Omni-MATH | 60.2 | Acc | 86.6 |
| HELM Safety - Simple Safety Tests | 100 | LM Evaluated Safety score | 86 |
| HELM Capabilities - MMLU-Pro | 84.3 | COT correct | 85.8 |
| HELM Safety - Bbq | 96.6 | BBQ accuracy | 79.7 |
| HELM Capabilities - WildBench | 83.81 | WB Score | 77.6 |
| Vals AI MATH 500 | 93.8 | Accuracy (%) | 75.7 |
Interactive version: theaggregate.ai/model?slug=claude-sonnet-4-20250514-thinking · How It Works · Data refreshed daily, snapshot 2026-10-20.