Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | DeepSeek-V3.1 | Gemini 2.5 Flash |
|---|---|---|
| GPQA Diamond | 74.9 | 82.8 |
| Humanity's Last Exam | 15.9 | 11 |
math
| Benchmark | DeepSeek-V3.1 | Gemini 2.5 Flash |
|---|---|---|
| AIME 2025 | 49.8 | 72 |
coding
| Benchmark | DeepSeek-V3.1 | Gemini 2.5 Flash |
|---|---|---|
| SWE-bench Verified | 66 | 60.4 |
| TerminalBench | 31.3 | — |
general
| Benchmark | DeepSeek-V3.1 | Gemini 2.5 Flash |
|---|---|---|
| SimpleQA | 93.4 | 26.9 |
agent
| Benchmark | DeepSeek-V3.1 | Gemini 2.5 Flash |
|---|---|---|
| BrowseComp | 30 | — |
multimodal
| Benchmark | DeepSeek-V3.1 | Gemini 2.5 Flash |
|---|---|---|
| MMMU | — | 79.7 |
Category Winners
coding
44.4
math
36.8
reasoning
46.4
general
98.3
multimodal
80.5
agent
4.4