| Benchmark | Claude Opus 4.7 | Nemotron 3 Super 120B A12B |
|---|---|---|
| AA Agentic Index | 64.6 | 8.8 |
| AA Intelligence | 55 | 25.7 |
| AA-LCR | 75.3 | 60.3 |
| AA-Omniscience | 27.3 | -41.5 |
| aime_2026 | 95.8 | 91.7 |
| Artificial Analysis Coding Index | 73.6 | 37.7 |
| browsecomp | 79.8 | 31.3 |
| critpt | 12 | 3.1 |
| gdpval | 58.6 | 9.9 |
| GPQA Diamond | 94.2 | 82.7 |
| HLE | 54.7 | 22.8 |
| hmmt_feb_2026 | 93.9 | 85.2 |
| IFBench | 58.6 | 72.6 |
| itbenchSre | 46.7 | 1.1 |
| OmniScience Accuracy | 48.9 | 24.3 |
| OmniScience Non-Hallucination | 57.7 | 13 |
| scicode | 54.5 | 42 |
| SWE-bench Multilingual | 80.5 | 45.8 |
| SWE-bench Verified | 87.6 | 53.7 |
| TauBench V3 - Banking | 34.6 | 10.3 |
| Terminal-Bench 2.0 | 69.4 | 31 |
| Terminal-Bench 2.1 | 83.1 | 38.6 |
| Terminal-Bench Hard | 54.5 | 28.8 |
| τ²-Bench Telecom (AA run) | 88.6 | 67.8 |
| τ³-Bench | 28.9 | 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.