Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | GPT-5.4 | Muse Spark |
|---|---|---|
| GPQA Diamond | 92.8 | 89.5 |
| Humanity's Last Exam | 39.8 | 58.4 |
| ARC-AGI v2 | 73.3 | 42.5 |
coding
| Benchmark | GPT-5.4 | Muse Spark |
|---|---|---|
| SWE-Bench Pro | 57.7 | 52.4 |
| SWE-bench Verified | — | 77.4 |
multimodal
| Benchmark | GPT-5.4 | Muse Spark |
|---|---|---|
| MMMU-Pro | 81.2 | 80.4 |
| CharXiv Reasoning | — | 86.4 |
| ScreenSpot Pro | — | 84.1 |
agent
| Benchmark | GPT-5.4 | Muse Spark |
|---|---|---|
| BrowseComp | 82.7 | — |
| Toolathlon | 54.6 | — |
| MCP Atlas | 67.2 | — |
math
| Benchmark | GPT-5.4 | Muse Spark |
|---|---|---|
| FrontierMath | 47.6 | — |
Category Winners
coding
M
Muse Spark52.6
math
69.1
reasoning
80.0
multimodal
M
Muse Spark83.3
agent
57.7