| Benchmark | Claude Fable 5 | Nemotron 3 Super 120B A12B |
|---|---|---|
| AA Agentic Index | 56.6 | 8.8 |
| AA Intelligence | 62.1 | 25.7 |
| AA-LCR | 76.7 | 60.3 |
| AA-Omniscience | 43.3 | -41.5 |
| Artificial Analysis Coding Index | 76.5 | 37.7 |
| browsecomp | 88 | 31.3 |
| critpt | 28.6 | 3.1 |
| gdpval | 61.9 | 9.9 |
| GPQA Diamond | 92.6 | 82.7 |
| HLE | 64.5 | 22.8 |
| IFBench | 63.5 | 72.6 |
| OmniScience Accuracy | 65.3 | 24.3 |
| OmniScience Non-Hallucination | 36.4 | 13 |
| scicode | 60.2 | 42 |
| SWE-bench Multilingual | 86.6 | 45.8 |
| SWE-bench Verified | 95 | 53.7 |
| TauBench V3 - Banking | 38.1 | 10.3 |
| Terminal-Bench 2.1 | 88 | 38.6 |
| Terminal-Bench Hard | 62.9 | 28.8 |
| τ²-Bench Telecom (AA run) | 98.5 | 67.8 |
| τ³-Bench | 26.8 | 10.1 |
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.