First Claude to match GPT-4, introduced the Opus/Sonnet/Haiku tier system
| Benchmark | Score | Rank |
|---|---|---|
HellaSwag Common sense reasoning about everyday situations | 95.4% | #11 / 36 |
ARC-C Grade-school science questions requiring reasoning | 96.4% | #25 / 40 |
MMLU Tests knowledge across 57 subjects from STEM to humanities | 86.8% | #36 / 53 |
HumanEval Coding ability - generating correct Python functions | 84.9% | #37 / 49 |
Arena Elo Human preference ranking via blind comparisons | 1248 | #37 / 41 |
MATH Competition-level mathematics problems | 60.1% | #40 / 49 |
GPQA PhD-level science questions even experts struggle with | 59.4% | #42 / 54 |