| Benchmark | Claude Fable 5 | GPT-4o mini |
|---|---|---|
| AA Agentic Index | 56.6 | 1 |
| AA Intelligence | 62.1 | 7 |
| ARC-AGI-2 | 76.8 | 0 |
| Artificial Analysis Coding Index | 76.5 | 11.4 |
| gdpval | 61.9 | 0 |
| GPQA Diamond | 92.6 | 42.6 |
| HLE | 64.5 | 4.2 |
| IFBench | 63.5 | 31 |
| MMMU-Pro | 84.2 | 41.5 |
| scicode | 60.2 | 22.9 |
| simplebench | 81.9 | 10.7 |
| SWE-bench Verified | 95 | 8.7 |
| TauBench V3 - Banking | 38.1 | 2.9 |
| Terminal-Bench 2.1 | 88 | 5.6 |
| τ³-Bench | 26.8 | 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.