Frontier reasoning model that preceded GPT-5.6, with strong coding, agentic, and ARC-AGI performance.
| Benchmark | Score | Rank |
|---|---|---|
MMMUvals.ai College-level multimodal reasoning across 30+ disciplines | 88.3% | #3 / 46 |
Terminal Agentic terminal coding tasks requiring multi-step execution | 85.6% | #3 / 56 |
ARC-AGIARC Prize Novel reasoning tasks requiring fluid intelligence | 90.6% | #3 / 26 |
GPQA PhD-level science questions even experts struggle with | 93.5% | #4 / 73 |
AA Coding Independent composite of implementation, terminal use, and real-codebase agent performance | 76.4% | #4 / 6 |
AA Intelligence Independent composite of agentic work, coding, scientific reasoning, and general capability | 54.8% | #5 / 6 |
hleArtificial Analysis | 43% | #5 / 61 |
MMLU-Provals.ai Harder 10-option successor to MMLU; more reasoning-focused | 88.1% | #12 / 45 |
OSWorld Computer use in real desktop environments | 47.5% | #12 / 13 |
LiveCodeBenchvals.ai Contamination-free competitive programming (filtered by cutoff date) | 85.3% | #17 / 48 |