Claude Sonnet 4 strikes a balance between thoughtful reasoning and practical responsiveness — it engages carefully with complex questions without the slower deliberation of heavier models. It handles nuanced writing, analysis, and coding with consistency, and tends to be direct while maintaining a measured, considered tone. A reliable mid-tier workhorse that doesn't sacrifice depth for speed.
| Benchmark | Score | Type | Recorded |
|---|---|---|---|
| AIME 2025 | 70.5 | accuracy | 5d ago |
| IFBench | 42.3 | prompt_level_loose_accuracy | 5d ago |
| TAU2 | 60.0 | accuracy | 5d ago |
| TerminalBench | 35.5 | accuracy | 5d ago |
| SWE-Bench | 72.7 | accuracy | 5d ago |
| LCR | 65.0 | pass@1_accuracy | 5d ago |
| Aider Polyglot | 61.3 | accuracy | 5d ago |