Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | DiffusionGemma 26B-A4B | MiniMax M1 80K |
|---|---|---|
| GPQA Diamond | 73.2 | 70 |
| Humanity's Last Exam | 11.9 | 8.4 |
multimodal
| Benchmark | DiffusionGemma 26B-A4B | MiniMax M1 80K |
|---|---|---|
| MMMU-Pro | 54.3 | — |
math
| Benchmark | DiffusionGemma 26B-A4B | MiniMax M1 80K |
|---|---|---|
| AIME 2025 | — | 76.9 |
coding
| Benchmark | DiffusionGemma 26B-A4B | MiniMax M1 80K |
|---|---|---|
| SWE-bench Verified | — | 56 |
agent
| Benchmark | DiffusionGemma 26B-A4B | MiniMax M1 80K |
|---|---|---|
| TAU-Bench Retail | — | 63.5 |
Category Winners
coding
29.0
math
45.8
reasoning
38.8
general
42.7
multimodal
15.6
agent
37.2