Highest GPQA Diamond score ever at 94.3%, doubled ARC-AGI-2 to 77.1%
| Benchmark | Score | Rank |
|---|---|---|
GPQA PhD-level science questions even experts struggle with | 94.3% | #1 / 54 |
Arena Elo Human preference ranking via blind comparisons | 1520 | #1 / 41 |
MMLU-Provals.ai Harder 10-option successor to MMLU; more reasoning-focused | 91% | #1 / 30 |
LiveCodeBenchvals.ai Contamination-free competitive programming (filtered by cutoff date) | 88.5% | #1 / 31 |
MMMUvals.ai College-level multimodal reasoning across 30+ disciplines | 88.2% | #1 / 33 |
ARC-AGI Novel reasoning tasks requiring fluid intelligence | 77.1% | #2 / 21 |
Terminal Agentic terminal coding tasks requiring multi-step execution | 68.5% | #3 / 37 |
SWE-bench Real-world GitHub issue resolution | 80.6% | #4 / 38 |
MMLU Tests knowledge across 57 subjects from STEM to humanities | 92.6% | #7 / 53 |
HumanEval Coding ability - generating correct Python functions | 94.2% | #14 / 49 |
MATH Competition-level mathematics problems | 95.1% | #16 / 49 |