Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | Gemma 4 26B-A4B | MiniMax M2.1 |
|---|---|---|
| GPQA Diamond | 82.3 | 81 |
| Humanity's Last Exam | 17.2 | 22 |
multimodal
| Benchmark | Gemma 4 26B-A4B | MiniMax M2.1 |
|---|---|---|
| MMMU-Pro | 73.8 | — |
math
| Benchmark | Gemma 4 26B-A4B | MiniMax M2.1 |
|---|---|---|
| AIME 2025 | — | 81 |
coding
| Benchmark | Gemma 4 26B-A4B | MiniMax M2.1 |
|---|---|---|
| SWE-bench Verified | — | 67 |
| TerminalBench | — | 47.9 |
| SciCode | — | 39 |
agent
| Benchmark | Gemma 4 26B-A4B | MiniMax M2.1 |
|---|---|---|
| BrowseComp | — | 62 |
| Toolathlon | — | 43.5 |
Category Winners
coding
58.3
math
53.6
reasoning
51.9
general
52.2
multimodal
66.5
agent
33.7