Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
coding
| Benchmark | MiniMax M3 | Qwen3.5-27B |
|---|---|---|
| SWE-bench Verified | 80.5 | 72.4 |
| SWE-Bench Pro | 59 | — |
multimodal
| Benchmark | MiniMax M3 | Qwen3.5-27B |
|---|---|---|
| MMMU-Pro | 78.1 | 75 |
| MMMU | — | 82.3 |
| CharXiv Reasoning | — | 79.5 |
| ScreenSpot Pro | — | 70.3 |
agent
| Benchmark | MiniMax M3 | Qwen3.5-27B |
|---|---|---|
| BrowseComp | 83.5 | 61 |
| MCP Atlas | 74.2 | — |
| APEX Agents | 27.7 | — |
reasoning
| Benchmark | MiniMax M3 | Qwen3.5-27B |
|---|---|---|
| GPQA Diamond | — | 85.5 |
| Humanity's Last Exam | — | 48.5 |
general
| Benchmark | MiniMax M3 | Qwen3.5-27B |
|---|---|---|
| MMMLU | — | 85.9 |
Category Winners
coding
68.3
reasoning
Q
Qwen3.5-27B83.6
general
Q
Qwen3.5-27B60.6
multimodal
77.2
agent
55.7