Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | DeepSeek-V3.2 (Thinking) | Grok-4 |
|---|---|---|
| GPQA Diamond | 82.4 | 87.5 |
| Humanity's Last Exam | 25.1 | 40 |
| ARC-AGI v2 | — | 15.9 |
math
| Benchmark | DeepSeek-V3.2 (Thinking) | Grok-4 |
|---|---|---|
| AIME 2025 | 93.1 | 91.7 |
coding
| Benchmark | DeepSeek-V3.2 (Thinking) | Grok-4 |
|---|---|---|
| SWE-bench Verified | 73.1 | — |
agent
| Benchmark | DeepSeek-V3.2 (Thinking) | Grok-4 |
|---|---|---|
| BrowseComp | 51.4 | — |
| Toolathlon | 35.2 | — |
Category Winners
coding
68.0
math
74.8
reasoning
57.7
agent
17.6