Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | Gemini 2.5 Flash-Lite | Qwen2.5 VL 72B Instruct |
|---|---|---|
| GPQA Diamond | 64.6 | — |
| Humanity's Last Exam | 5.1 | — |
math
| Benchmark | Gemini 2.5 Flash-Lite | Qwen2.5 VL 72B Instruct |
|---|---|---|
| AIME 2025 | 49.8 | — |
coding
| Benchmark | Gemini 2.5 Flash-Lite | Qwen2.5 VL 72B Instruct |
|---|---|---|
| SWE-bench Verified | 31.6 | — |
multimodal
| Benchmark | Gemini 2.5 Flash-Lite | Qwen2.5 VL 72B Instruct |
|---|---|---|
| MMMU | 72.9 | 70.2 |
| MMMU-Pro | — | 51.1 |
| ScreenSpot Pro | — | 43.6 |
general
| Benchmark | Gemini 2.5 Flash-Lite | Qwen2.5 VL 72B Instruct |
|---|---|---|
| SimpleQA | 10.7 | — |
agent
| Benchmark | Gemini 2.5 Flash-Lite | Qwen2.5 VL 72B Instruct |
|---|---|---|
| OSWorld | — | 8.8 |
Category Winners
coding
2.2
math
8.1
reasoning
26.8
general
14.4
multimodal
64.8
agent
Q
Qwen2.5 VL 72B Instruct4.3