First model to break 1500 Arena Elo, 100% on AIME with code execution
| Benchmark | Score | Rank |
|---|---|---|
MMLU Tests knowledge across 57 subjects from STEM to humanities | 93.2% | #2 / 53 |
ARC-C Grade-school science questions requiring reasoning | 98.8% | #2 / 40 |
HellaSwag Common sense reasoning about everyday situations | 97.9% | #2 / 36 |
MMLU-Provals.ai Harder 10-option successor to MMLU; more reasoning-focused | 90.1% | #2 / 30 |
Arena Elo Human preference ranking via blind comparisons | 1501 | #3 / 41 |
FrontierMath Unpublished research-level mathematics problems | 38% | #3 / 4 |
MMMUvals.ai College-level multimodal reasoning across 30+ disciplines | 87.5% | #3 / 33 |
HumanEval Coding ability - generating correct Python functions | 96.2% | #4 / 49 |
MATH Competition-level mathematics problems | 100% | #4 / 49 |
LiveCodeBenchvals.ai Contamination-free competitive programming (filtered by cutoff date) | 86.4% | #5 / 31 |
GPQA PhD-level science questions even experts struggle with | 91.9% | #6 / 54 |
ARC-AGI Novel reasoning tasks requiring fluid intelligence | 45.1% | #9 / 21 |
SWE-bench Real-world GitHub issue resolution | 76.2% | #13 / 38 |
TerminalArtificial Analysis Agentic terminal coding tasks requiring multi-step execution | 41.7% | #14 / 37 |