Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | o3-mini | Qwen3 VL 4B Instruct |
|---|---|---|
| GPQA Diamond | 77.2 | — |
coding
| Benchmark | o3-mini | Qwen3 VL 4B Instruct |
|---|---|---|
| SWE-bench Verified | 49.3 | — |
general
| Benchmark | o3-mini | Qwen3 VL 4B Instruct |
|---|---|---|
| SimpleQA | 15 | 48 |
agent
| Benchmark | o3-mini | Qwen3 VL 4B Instruct |
|---|---|---|
| TAU-Bench Retail | 57.6 | — |
| OSWorld | — | 26.2 |
math
| Benchmark | o3-mini | Qwen3 VL 4B Instruct |
|---|---|---|
| FrontierMath | 9.2 | — |
| AIME 2025 | — | 46.6 |
multimodal
| Benchmark | o3-mini | Qwen3 VL 4B Instruct |
|---|---|---|
| MMMU-Pro | — | 53.2 |
| CharXiv Reasoning | — | 39.7 |
| ScreenSpot Pro | — | 59.5 |
Category Winners
coding
15.4
math
16.1
reasoning
67.7
general
Q
Qwen3 VL 4B Instruct65.2
multimodal
Q
Qwen3 VL 4B Instruct16.9
agent
21.9