First public Mythos-class model, a tier above Opus, with strong reasoning, coding, and long-horizon professional-work performance.
| Benchmark | Score | Rank |
|---|---|---|
LiveCodeBenchvals.ai Contamination-free competitive programming (filtered by cutoff date) | 89.8% | #1 / 53 |
hleArtificial Analysis | 53.3% | #1 / 66 |
MMLU-Provals.ai Harder 10-option successor to MMLU; more reasoning-focused | 91.5% | #2 / 51 |
AA Intelligence Independent composite of agentic work, coding, scientific reasoning, and general capability | 59.9% | #2 / 11 |
MMMUvals.ai College-level multimodal reasoning across 30+ disciplines | 89.3% | #2 / 51 |
AA Coding Independent composite of implementation, terminal use, and real-codebase agent performance | 77.2% | #4 / 11 |
Terminal Agentic terminal coding tasks requiring multi-step execution | 83.1% | #5 / 56 |
GPQAArtificial Analysis PhD-level science questions even experts struggle with | 92.6% | #13 / 78 |