Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | DeepSeek-V3.1 | Nemotron 3 Super (120B A12B) |
|---|---|---|
| GPQA Diamond | 74.9 | 82.7 |
| Humanity's Last Exam | 15.9 | 22.8 |
math
| Benchmark | DeepSeek-V3.1 | Nemotron 3 Super (120B A12B) |
|---|---|---|
| AIME 2025 | 49.8 | 90.2 |
coding
| Benchmark | DeepSeek-V3.1 | Nemotron 3 Super (120B A12B) |
|---|---|---|
| SWE-bench Verified | 66 | 53.7 |
| TerminalBench | 31.3 | 25.8 |
| SciCode | — | 42.0 |
general
| Benchmark | DeepSeek-V3.1 | Nemotron 3 Super (120B A12B) |
|---|---|---|
| SimpleQA | 93.4 | — |
agent
| Benchmark | DeepSeek-V3.1 | Nemotron 3 Super (120B A12B) |
|---|---|---|
| BrowseComp | 30 | 31.3 |
Category Winners
coding
44.1
math
70.5
reasoning
55.0
general
98.5
agent
5.1