| Benchmark | Claude Fable 5 | o1 |
|---|---|---|
| AA Agentic Index | 56.6 | 31.1 |
| AA Intelligence | 62.1 | 24 |
| AA-LCR | 76.7 | 63.3 |
| AA-Omniscience | 43.3 | -10.6 |
| arena_vision | 1307 | 1193 |
| Artificial Analysis Coding Index | 76.5 | 39.7 |
| critpt | 28.6 | 0.3 |
| gdpval | 61.9 | 11.5 |
| GPQA Diamond | 92.6 | 75.7 |
| HLE | 64.5 | 7.7 |
| IFBench | 63.5 | 70.3 |
| livebench | 78.3 | 52.3 |
| OmniScience Accuracy | 65.3 | 34.7 |
| OmniScience Non-Hallucination | 36.4 | 30.7 |
| scicode | 60.2 | 35.8 |
| simplebench | 81.9 | 41.7 |
| strongreject | 98.7 | 97 |
| SWE-bench Verified | 95 | 48.9 |
| Terminal-Bench Hard | 62.9 | 12.9 |
| τ²-Bench Telecom (AA run) | 98.5 | 62.6 |
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.