Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | GPT-5.4 | Qwen3.5-397B-A17B |
|---|---|---|
| GPQA Diamond | 92.8 | 88.4 |
| Humanity's Last Exam | 39.8 | 28.7 |
| ARC-AGI v2 | 73.3 | — |
coding
| Benchmark | GPT-5.4 | Qwen3.5-397B-A17B |
|---|---|---|
| SWE-Bench Pro | 57.7 | — |
| SWE-bench Verified | — | 76.4 |
multimodal
| Benchmark | GPT-5.4 | Qwen3.5-397B-A17B |
|---|---|---|
| MMMU-Pro | 81.2 | — |
agent
| Benchmark | GPT-5.4 | Qwen3.5-397B-A17B |
|---|---|---|
| BrowseComp | 82.7 | 69 |
| Toolathlon | 54.6 | 38.3 |
| MCP Atlas | 67.2 | — |
math
| Benchmark | GPT-5.4 | Qwen3.5-397B-A17B |
|---|---|---|
| FrontierMath | 47.6 | — |
general
| Benchmark | GPT-5.4 | Qwen3.5-397B-A17B |
|---|---|---|
| MMMLU | — | 88.5 |
Category Winners
coding
Q
Qwen3.5-397B-A17B74.7
math
69.1
reasoning
81.8
general
Q
Qwen3.5-397B-A17B70.4
multimodal
83.1
agent
63.1