| Benchmark | Gemini 1.5 Pro | Nemotron 3 Super 120B A12B |
|---|---|---|
| AA Intelligence | 10 | 25.7 |
| Artificial Analysis Coding Index | 23.6 | 37.7 |
| GPQA Diamond | 59.1 | 82.7 |
| GSM8K | 95.2 | 90.7 |
| hellaswag | 93.3 | 89 |
| HLE | 4.9 | 22.8 |
| humaneval | 86.6 | 80.5 |
| LiveCodeBench | 30.5 | 82.1 |
| mmlu | 86.8 | 86 |
| MMLU-Pro | 75.8 | 83.8 |
| OpenBookQA | 95.2 | 48.6 |
| RULER 1M | 85 | 93.9 |
| RULER 256K | 91.6 | 96.7 |
| RULER 512K | 86.1 | 95.7 |
| scicode | 29.5 | 42 |
| SWE-bench Verified | 34.2 | 53.7 |
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.