Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | Gemma 4 31B | Nova 2 Pro |
|---|---|---|
| GPQA Diamond | 84.3 | 81.4 |
| Humanity's Last Exam | 26.5 | — |
multimodal
| Benchmark | Gemma 4 31B | Nova 2 Pro |
|---|---|---|
| MMMU-Pro | 76.9 | 63.5 |
math
| Benchmark | Gemma 4 31B | Nova 2 Pro |
|---|---|---|
| AIME 2025 | — | 92.3 |
coding
| Benchmark | Gemma 4 31B | Nova 2 Pro |
|---|---|---|
| SWE-bench Verified | — | 70 |
| TerminalBench | — | 41.3 |
Category Winners
coding
Nova 2 Pro66.8
math
Nova 2 Pro74.0
reasoning
Nova 2 Pro74.2
general
71.5
multimodal
73.4