Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | Claude 3.7 Sonnet | Qwen3 VL 32B Thinking |
|---|---|---|
| GPQA Diamond | 84.8 | 73.1 |
math
| Benchmark | Claude 3.7 Sonnet | Qwen3 VL 32B Thinking |
|---|---|---|
| AIME 2025 | 54.8 | 83.7 |
coding
| Benchmark | Claude 3.7 Sonnet | Qwen3 VL 32B Thinking |
|---|---|---|
| SWE-bench Verified | 70.3 | — |
| TerminalBench | 35.2 | — |
multimodal
| Benchmark | Claude 3.7 Sonnet | Qwen3 VL 32B Thinking |
|---|---|---|
| MMMU | 75 | — |
| MMMU-Pro | — | 68.1 |
| CharXiv Reasoning | — | 65.2 |
| ScreenSpot Pro | — | 57.1 |
agent
| Benchmark | Claude 3.7 Sonnet | Qwen3 VL 32B Thinking |
|---|---|---|
| TAU-Bench Retail | 81.2 | — |
| OSWorld | — | 41 |
Category Winners
coding
56.8
math
Q
Qwen3 VL 32B Thinking58.7
reasoning
81.8
general
Q
Qwen3 VL 32B Thinking74.7
multimodal
70.1
agent
84.7