Built-in thinking mode, native audio and video understanding
| Benchmark | Score | Rank |
|---|---|---|
HellaSwag Common sense reasoning about everyday situations | 96.2% | #10 / 36 |
ARC-C Grade-school science questions requiring reasoning | 97.8% | #12 / 40 |
MMMUvals.ai College-level multimodal reasoning across 30+ disciplines | 81.3% | #15 / 39 |
MMLU Tests knowledge across 57 subjects from STEM to humanities | 90.8% | #19 / 54 |
Arena Elo Human preference ranking via blind comparisons | 1451 | #21 / 52 |
HumanEval Coding ability - generating correct Python functions | 93.2% | #22 / 50 |
MATH Competition-level mathematics problems | 89.5% | #25 / 50 |
MMLU-Provals.ai Harder 10-option successor to MMLU; more reasoning-focused | 84.1% | #26 / 38 |
LiveCodeBenchvals.ai Contamination-free competitive programming (filtered by cutoff date) | 79.2% | #26 / 40 |
SWE-bench Real-world GitHub issue resolution | 59.6% | #32 / 40 |
GPQA PhD-level science questions even experts struggle with | 84% | #33 / 64 |
TerminalArtificial Analysis Agentic terminal coding tasks requiring multi-step execution | 26.5% | #37 / 48 |