| Benchmark | Claude Opus 4.6 | Nemotron 3 Super 120B A12B |
|---|---|---|
| AA Agentic Index | 67.6 | 8.8 |
| AA Intelligence | 44.9 | 25.7 |
| AA-LCR | 74.3 | 60.3 |
| AA-Omniscience | 13.7 | -41.5 |
| AIME 2025 | 99.8 | 90.2 |
| aime_2026 | 96.7 | 91.7 |
| AIME25 no tools | 95.6 | 92.2 |
| apexAgents | 33 | 1.8 |
| Artificial Analysis Coding Index | 48.1 | 37.7 |
| browsecomp | 86.8 | 31.3 |
| critpt | 12.6 | 3.1 |
| gdpval | 55.9 | 9.9 |
| GPQA Diamond | 91.3 | 82.7 |
| GSM8K | 0.6 | 90.7 |
| HLE | 53.1 | 22.8 |
| hmmt_feb_2026 | 96.2 | 85.2 |
| IFBench | 62.5 | 72.6 |
| LiveCodeBench | 88.8 | 82.1 |
| MMLU-Pro | 89.1 | 83.8 |
| multichallenge | 56 | 55.2 |
| OmniScience Accuracy | 47 | 24.3 |
| OmniScience Non-Hallucination | 37.2 | 13 |
| scicode | 52 | 42 |
| SWE-bench Multilingual | 77.8 | 45.8 |
| SWE-bench Verified | 80.8 | 53.7 |
| Tau2 retail | 91.9 | 62.8 |
| Terminal-Bench 2.0 | 65.4 | 31 |
| Terminal-Bench Hard | 65.4 | 28.8 |
| τ²-Bench Telecom (AA run) | 99.3 | 67.8 |
| τ³-Bench | 72.4 | 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.