Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | DeepSeek-V4-Flash-Max | Qwen3 VL 32B Thinking |
|---|---|---|
| GPQA Diamond | 88.1 | 73.1 |
| Humanity's Last Exam | 45.1 | — |
coding
| Benchmark | DeepSeek-V4-Flash-Max | Qwen3 VL 32B Thinking |
|---|---|---|
| SWE-bench Verified | 79 | — |
| SWE-Bench Pro | 52.6 | — |
general
| Benchmark | DeepSeek-V4-Flash-Max | Qwen3 VL 32B Thinking |
|---|---|---|
| SimpleQA | 34.1 | 55.4 |
agent
| Benchmark | DeepSeek-V4-Flash-Max | Qwen3 VL 32B Thinking |
|---|---|---|
| BrowseComp | 73.2 | — |
| Toolathlon | 47.8 | — |
| MCP Atlas | 69 | — |
| OSWorld | — | 41 |
math
| Benchmark | DeepSeek-V4-Flash-Max | Qwen3 VL 32B Thinking |
|---|---|---|
| AIME 2025 | — | 83.7 |
multimodal
| Benchmark | DeepSeek-V4-Flash-Max | Qwen3 VL 32B Thinking |
|---|---|---|
| MMMU-Pro | — | 68.1 |
| CharXiv Reasoning | — | 65.2 |
| ScreenSpot Pro | — | 57.1 |
Category Winners
coding
49.7
math
Q
Qwen3 VL 32B Thinking58.7
reasoning
82.0
general
Q
Qwen3 VL 32B Thinking74.7
multimodal
Q
Qwen3 VL 32B Thinking37.8
agent
52.0