Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | DeepSeek-V3.2-Exp | Grok-4 |
|---|---|---|
| GPQA Diamond | 79.9 | 87.5 |
| Humanity's Last Exam | 19.8 | 40 |
| ARC-AGI v2 | — | 15.9 |
math
| Benchmark | DeepSeek-V3.2-Exp | Grok-4 |
|---|---|---|
| AIME 2025 | 89.3 | 91.7 |
coding
| Benchmark | DeepSeek-V3.2-Exp | Grok-4 |
|---|---|---|
| SWE-bench Verified | 67.8 | — |
| TerminalBench | 37.7 | — |
general
| Benchmark | DeepSeek-V3.2-Exp | Grok-4 |
|---|---|---|
| SimpleQA | 97.1 | — |
agent
| Benchmark | DeepSeek-V3.2-Exp | Grok-4 |
|---|---|---|
| BrowseComp | 40.1 | — |
Category Winners
coding
58.6
math
X
Grok-472.6
reasoning
X
Grok-455.7
general
98.8
agent
11.0