| Benchmark | Claude Opus 4.6 | GPT-4o mini |
|---|---|---|
| AA Agentic Index | 67.6 | 1 |
| AA Intelligence | 44.9 | 7 |
| ARC-AGI-2 | 68.8 | 0 |
| Artificial Analysis Coding Index | 48.1 | 11.4 |
| Fortress | 20.5 | 48.1 |
| gdpval | 55.9 | 0 |
| GPQA Diamond | 91.3 | 42.6 |
| GSM8K | 0.6 | 84.3 |
| HLE | 53.1 | 4.2 |
| IFBench | 62.5 | 31 |
| LiveCodeBench | 88.8 | 35.5 |
| MMMU-Pro | 77.3 | 41.5 |
| scicode | 52 | 22.9 |
| simplebench | 67.6 | 10.7 |
| SWE-bench Verified | 80.8 | 8.7 |
| τ³-Bench | 72.4 | 2.9 |
Best tracked score per model per benchmark (default configuration; source-attributed). ↓ marks lower-is-better metrics. Open either model for its full surface, provenance and pricing.