Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | GPT-5 mini | Phi 4 Reasoning Plus |
|---|---|---|
| GPQA Diamond | 82.3 | 68.9 |
| Humanity's Last Exam | 16.7 | — |
math
| Benchmark | GPT-5 mini | Phi 4 Reasoning Plus |
|---|---|---|
| AIME 2025 | 91.1 | 78 |
| FrontierMath | 22.1 | — |
Category Winners
math
51.5
reasoning
52.5