Neura Intelligence Index
AI Model Benchmarks
| Benchmark | DeepSeek-V4.1-Flash-1017 | Gemini 3 Pro |
|---|---|---|
| Humanity's Last Exam | 42.6 | 45.8 |
| GPQA Diamond | — | 91.9 |
| ARC-AGI v2 | — | 31.1 |
| Benchmark | DeepSeek-V4.1-Flash-1017 | Gemini 3 Pro |
|---|---|---|
| AIME 2025 | — | 100 |
| Benchmark | DeepSeek-V4.1-Flash-1017 | Gemini 3 Pro |
|---|---|---|
| SWE-bench Verified | — | 76.2 |
| Benchmark | DeepSeek-V4.1-Flash-1017 | Gemini 3 Pro |
|---|---|---|
| MMMU-Pro | — | 81 |
| CharXiv Reasoning | — | 81.4 |
| ScreenSpot Pro | — | 72.7 |