Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | GPT-5.5 | Hy3 |
|---|---|---|
| GPQA Diamond | 93.6 | 90.4 |
| Humanity's Last Exam | 52.2 | — |
| ARC-AGI v2 | 85 | — |
coding
| Benchmark | GPT-5.5 | Hy3 |
|---|---|---|
| SWE-Bench Pro | 58.6 | 57.9 |
| SWE-bench Verified | — | 78 |
multimodal
| Benchmark | GPT-5.5 | Hy3 |
|---|---|---|
| MMMU-Pro | 83.2 | — |
agent
| Benchmark | GPT-5.5 | Hy3 |
|---|---|---|
| BrowseComp | 84.4 | 84.2 |
| Toolathlon | 55.6 | 48.5 |
| MCP Atlas | 75.3 | 79.1 |
| APEX Agents | — | 25.6 |
math
| Benchmark | GPT-5.5 | Hy3 |
|---|---|---|
| FrontierMath | 35.4 | — |
general
| Benchmark | GPT-5.5 | Hy3 |
|---|---|---|
| MRCR v2 | 74 | — |
Category Winners
coding
T
Hy363.0
math
51.0
reasoning
91.3
general
89.3
multimodal
86.9
agent
74.9