Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
coding
| Benchmark | Gemini 3.6 Flash | GPT-5 |
|---|---|---|
| SWE-Bench Pro | 58.7 | — |
| SWE-bench Verified | — | 74.9 |
multimodal
| Benchmark | Gemini 3.6 Flash | GPT-5 |
|---|---|---|
| CharXiv Reasoning | 89.4 | 81.1 |
| MMMU | — | 84.2 |
| MMMU-Pro | — | 78.4 |
general
| Benchmark | Gemini 3.6 Flash | GPT-5 |
|---|---|---|
| MRCR v2 | 54 | — |
reasoning
| Benchmark | Gemini 3.6 Flash | GPT-5 |
|---|---|---|
| GPQA Diamond | — | 85.7 |
| Humanity's Last Exam | — | 24.8 |
math
| Benchmark | Gemini 3.6 Flash | GPT-5 |
|---|---|---|
| AIME 2025 | — | 94.6 |
| FrontierMath | — | 26.3 |
agent
| Benchmark | Gemini 3.6 Flash | GPT-5 |
|---|---|---|
| BrowseComp | — | 54.9 |
Category Winners
coding
71.5
math
57.2
reasoning
57.9
general
56.1
multimodal
84.7
agent
30.4