Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | Nemotron 3 Super (120B A12B) | o1 |
|---|---|---|
| GPQA Diamond | 82.7 | 78 |
| Humanity's Last Exam | 22.8 | — |
math
| Benchmark | Nemotron 3 Super (120B A12B) | o1 |
|---|---|---|
| AIME 2025 | 90.2 | — |
| FrontierMath | — | 5.5 |
coding
| Benchmark | Nemotron 3 Super (120B A12B) | o1 |
|---|---|---|
| SWE-bench Verified | 53.7 | 41 |
| TerminalBench | 25.8 | — |
| SciCode | 42.0 | — |
agent
| Benchmark | Nemotron 3 Super (120B A12B) | o1 |
|---|---|---|
| BrowseComp | 31.3 | — |
| TAU-Bench Retail | — | 70.8 |
multimodal
| Benchmark | Nemotron 3 Super (120B A12B) | o1 |
|---|---|---|
| MMMU | — | 77.6 |
Category Winners
coding
26.3
math
70.2
reasoning
71.5
general
65.4
multimodal
76.2
agent
59.1