| Benchmark | Nemotron 3 Super 120B A12B | o4-mini |
|---|---|---|
| AA Agentic Index | 8.8 | 36.1 |
| AA Intelligence | 25.7 | 26.1 |
| AA-LCR | 60.3 | 60 |
| AA-Omniscience | -41.5 | -35.7 |
| AIME 2025 | 90.2 | 92.7 |
| Artificial Analysis Coding Index | 37.7 | 25.6 |
| browsecomp | 31.3 | 51.5 |
| critpt | 3.1 | 0.6 |
| gdpval | 9.9 | 25.4 |
| GPQA Diamond | 82.7 | 81.4 |
| GSM8K | 90.7 | 0 |
| HLE | 22.8 | 16.5 |
| IFBench | 72.6 | 68.7 |
| LiveCodeBench | 82.1 | 84.5 |
| mmlu_prox | 79.4 | 69.3 |
| multichallenge | 55.2 | 44.9 |
| OmniScience Accuracy | 24.3 | 24.8 |
| OmniScience Non-Hallucination | 13 | 19.5 |
| scicode | 42 | 46.5 |
| SWE-bench Verified | 53.7 | 68.1 |
| Terminal-Bench Hard | 28.8 | 15.2 |
| τ²-Bench Telecom (AA run) | 67.8 | 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.