Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | Gemma 4 26B-A4B | Nova 2 Lite |
|---|---|---|
| GPQA Diamond | 82.3 | 79.6 |
| Humanity's Last Exam | 17.2 | — |
multimodal
| Benchmark | Gemma 4 26B-A4B | Nova 2 Lite |
|---|---|---|
| MMMU-Pro | 73.8 | 61.8 |
math
| Benchmark | Gemma 4 26B-A4B | Nova 2 Lite |
|---|---|---|
| AIME 2025 | — | 91 |
coding
| Benchmark | Gemma 4 26B-A4B | Nova 2 Lite |
|---|---|---|
| SWE-bench Verified | — | 64.5 |
| TerminalBench | — | 32.5 |
agent
| Benchmark | Gemma 4 26B-A4B | Nova 2 Lite |
|---|---|---|
| MCP Atlas | — | 24.6 |
Category Winners
coding
Nova 2 Lite44.1
math
Nova 2 Lite71.6
reasoning
Nova 2 Lite71.9
general
52.2
multimodal
65.6
agent
Nova 2 Lite0.0