| Benchmark | Nemotron 3 Super 120B A12B | Qwen3.6 35B A3B |
|---|---|---|
| AA Agentic Index | 8.8 | 21.6 |
| AA Intelligence | 25.7 | 32.1 |
| AA-LCR | 60.3 | 66.7 |
| AA-Omniscience | -41.5 | -22.2 |
| aime_2026 | 91.7 | 92.7 |
| Artificial Analysis Coding Index | 37.7 | 41.9 |
| browsecomp | 31.3 | 48.7 |
| critpt | 3.1 | 0.3 |
| gdpval | 9.9 | 27.8 |
| GPQA Diamond | 82.7 | 86 |
| HLE | 22.8 | 22.2 |
| HMMT 2025 | 94.7 | 89.1 |
| hmmt_feb_2026 | 85.2 | 83.6 |
| IFBench | 72.6 | 64.4 |
| MMLU-Pro | 83.8 | 85.6 |
| OmniScience Accuracy | 24.3 | 18.8 |
| OmniScience Non-Hallucination | 13 | 49.5 |
| scicode | 42 | 35.8 |
| SWE-bench Multilingual | 45.8 | 67.2 |
| SWE-bench Verified | 53.7 | 73.4 |
| TauBench V3 - Banking | 10.3 | 9.3 |
| Terminal-Bench 2.0 | 31 | 51.5 |
| Terminal-Bench 2.1 | 38.6 | 44.9 |
| Terminal-Bench Hard | 28.8 | 34.8 |
| τ²-Bench Telecom (AA run) | 67.8 | 95.3 |
| τ³-Bench | 10.1 | 67.2 |
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.