| Benchmark | GPT-5.1 | Nemotron 3 Super 120B A12B |
|---|---|---|
| AA Agentic Index | 32.2 | 8.8 |
| AA Intelligence | 37.5 | 25.7 |
| AA-LCR | 76.7 | 60.3 |
| AA-Omniscience | 5.4 | -41.5 |
| AIME 2025 | 94.2 | 90.2 |
| Artificial Analysis Coding Index | 49.4 | 37.7 |
| browsecomp | 50.8 | 31.3 |
| critpt | 4.9 | 3.1 |
| gdpval | 25 | 9.9 |
| GPQA Diamond | 88.1 | 82.7 |
| HLE | 28.5 | 22.8 |
| IFBench | 72.9 | 72.6 |
| MMLU-Pro | 87 | 83.8 |
| multichallenge | 63.4 | 55.2 |
| OmniScience Accuracy | 37.7 | 24.3 |
| OmniScience Non-Hallucination | 48.1 | 13 |
| scicode | 43.3 | 42 |
| SWE-bench Verified | 76.3 | 53.7 |
| Tau2 airline | 67 | 56.3 |
| Tau2 retail | 77.9 | 62.8 |
| TauBench V3 - Banking | 15.9 | 10.3 |
| Terminal-Bench 2.0 | 47.6 | 31 |
| Terminal-Bench 2.1 | 52.4 | 38.6 |
| Terminal-Bench Hard | 45.5 | 28.8 |
| τ²-Bench Telecom (AA run) | 81.9 | 67.8 |
| τ³-Bench | 14 | 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.