| Benchmark | Claude Opus 4.8 | GPT-4o mini |
|---|---|---|
| AA Agentic Index | 49.4 | 1 |
| AA Intelligence | 57.3 | 7 |
| ARC-AGI-2 | 72.1 | 0 |
| Artificial Analysis Coding Index | 74.3 | 11.4 |
| Fortress | 18.2 | 48.1 |
| gdpval | 54.2 | 0 |
| GPQA Diamond | 93.6 | 42.6 |
| HLE | 57.9 | 4.2 |
| IFBench | 62.2 | 31 |
| MMMU-Pro | 78.9 | 41.5 |
| scicode | 53.5 | 22.9 |
| simplebench | 64.8 | 10.7 |
| SWE-bench Verified | 88.6 | 8.7 |
| TauBench V3 - Banking | 34.2 | 2.9 |
| Terminal-Bench 2.1 | 85 | 5.6 |
| Video-MME | 86 | 68.9 |
| τ³-Bench | 27.6 | 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.