| Benchmark | Claude Fable 5 | o4-mini |
|---|---|---|
| AA Agentic Index | 56.6 | 36.1 |
| AA Intelligence | 62.1 | 26.1 |
| AA-LCR | 76.7 | 60 |
| AA-Omniscience | 43.3 | -35.7 |
| arc_agi_1 | 90.5 | 58.7 |
| ARC-AGI-2 | 76.8 | 6.1 |
| arena_vision | 1307 | 1201 |
| Artificial Analysis Coding Index | 76.5 | 25.6 |
| browsecomp | 88 | 51.5 |
| critpt | 28.6 | 0.6 |
| gdpval | 61.9 | 25.4 |
| GPQA Diamond | 92.6 | 81.4 |
| HLE | 64.5 | 16.5 |
| IFBench | 63.5 | 68.7 |
| MMMU-Pro | 84.2 | 69.2 |
| OmniScience Accuracy | 65.3 | 24.8 |
| OmniScience Non-Hallucination | 36.4 | 19.5 |
| scicode | 60.2 | 46.5 |
| simplebench | 81.9 | 38.7 |
| simpleqa_verified | 68.3 | 23.9 |
| strongreject | 98.7 | 96 |
| SWE-bench Verified | 95 | 68.1 |
| Terminal-Bench Hard | 62.9 | 15.2 |
| τ²-Bench Telecom (AA run) | 98.5 | 55.6 |
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.