Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | Claude Sonnet 4.5 | Grok-3 |
|---|---|---|
| GPQA Diamond | 83.4 | 84.6 |
math
| Benchmark | Claude Sonnet 4.5 | Grok-3 |
|---|---|---|
| AIME 2025 | 87 | 93.3 |
general
| Benchmark | Claude Sonnet 4.5 | Grok-3 |
|---|---|---|
| MMMLU | 89.1 | — |
agent
| Benchmark | Claude Sonnet 4.5 | Grok-3 |
|---|---|---|
| OSWorld | 61.4 | — |
| TAU-Bench Retail | 86.2 | — |
coding
| Benchmark | Claude Sonnet 4.5 | Grok-3 |
|---|---|---|
| TerminalBench | 50 | — |
multimodal
| Benchmark | Claude Sonnet 4.5 | Grok-3 |
|---|---|---|
| MMMU | — | 78 |
Category Winners
coding
92.8
math
X
Grok-375.0
reasoning
X
Grok-381.3
general
72.5
multimodal
X
Grok-377.0
agent
84.0