Claude Sonnet 4
Anthropic·
LLMsmidcloud
Overview
Next-generation Sonnet with improved instruction following and coding.
Capabilities and innovations
Improved instruction followingEnhanced codingBetter reasoningImproved AlignmentBetter Calibration
Benchmarks
| Benchmark | Score | Source |
|---|---|---|
| GPQA Diamond | 86.4% | unsourced |
| Humanity’s Last Exam (no tools) | 30.2% | unsourced |
| SWE-bench Verified | 62.5% | unsourced |
| MMLU | 91.2% | unsourced |
| MMLU-Pro | 83.86% | unsourced |
- SWE-bench Pro: benchmark did not exist at release
- Terminal-Bench 2.x: benchmark did not exist at release
- MMMU-Pro: benchmark did not exist at release
Links
More from Anthropic
Data curated by AI Model Timeline. See the methodology for admission criteria, benchmark eras and source priority.