| Benchmark | Claude Opus 4.5 | Nemotron 3 Super 120B A12B |
|---|---|---|
| AA Agentic Index | 59.6 | 8.8 |
| AA Intelligence | 41.9 | 25.7 |
| AA-LCR | 76 | 60.3 |
| AA-Omniscience | 14 | -41.5 |
| AIME 2025 | 92.8 | 90.2 |
| aime_2026 | 95.1 | 91.7 |
| AIME25 no tools | 91 | 92.2 |
| Artificial Analysis Coding Index | 47.8 | 37.7 |
| browsecomp | 37 | 31.3 |
| critpt | 4.6 | 3.1 |
| gdpval | 47.3 | 9.9 |
| GPQA Diamond | 87 | 82.7 |
| HLE | 30.8 | 22.8 |
| hmmt_feb_2026 | 85.3 | 85.2 |
| IFBench | 58 | 72.6 |
| LiveCodeBench | 87 | 82.1 |
| MMLU-Pro | 90 | 83.8 |
| multichallenge | 59 | 55.2 |
| OmniScience Accuracy | 46.6 | 24.3 |
| OmniScience Non-Hallucination | 39 | 13 |
| scicode | 50 | 42 |
| SWE-bench Multilingual | 77.5 | 45.8 |
| SWE-bench Verified | 81.5 | 53.7 |
| Tau2 retail | 88.9 | 62.8 |
| Terminal-Bench 2.0 | 59.3 | 31 |
| Terminal-Bench Hard | 47 | 28.8 |
| τ²-Bench Telecom (AA run) | 89.5 | 67.8 |
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.