Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | Gemma 4 31B | OpenAI Codex |
|---|---|---|
| GPQA Diamond | 84.3 | — |
| Humanity's Last Exam | 26.5 | — |
multimodal
| Benchmark | Gemma 4 31B | OpenAI Codex |
|---|---|---|
| MMMU-Pro | 76.9 | — |
coding
| Benchmark | Gemma 4 31B | OpenAI Codex |
|---|---|---|
| SWE-bench Verified | — | 69.1 |
Category Winners
coding
59.1
reasoning
61.2
general
76.3
multimodal
74.5
agent
78.2