Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | GPT-5.6 Sol | Grok-4 Heavy |
|---|---|---|
| GPQA Diamond | 94.6 | 88.4 |
| Humanity's Last Exam | — | 50.7 |
coding
| Benchmark | GPT-5.6 Sol | Grok-4 Heavy |
|---|---|---|
| SWE-Bench Pro | 64.6 | — |
multimodal
| Benchmark | GPT-5.6 Sol | Grok-4 Heavy |
|---|---|---|
| MMMU-Pro | 83 | — |
agent
| Benchmark | GPT-5.6 Sol | Grok-4 Heavy |
|---|---|---|
| BrowseComp | 90.4 | — |
| Toolathlon | 58 | — |
math
| Benchmark | GPT-5.6 Sol | Grok-4 Heavy |
|---|---|---|
| FrontierMath | 89 | — |
| AIME 2025 | — | 100 |
general
| Benchmark | GPT-5.6 Sol | Grok-4 Heavy |
|---|---|---|
| MRCR v2 | 91.5 | — |
Category Winners
coding
74.4
math
98.2
reasoning
91.0
general
93.3
multimodal
86.3
agent
77.0