Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | GPT-5.4 nano | Phi 4 Reasoning |
|---|---|---|
| GPQA Diamond | 82.8 | 65.8 |
| Humanity's Last Exam | 24.3 | — |
coding
| Benchmark | GPT-5.4 nano | Phi 4 Reasoning |
|---|---|---|
| SWE-Bench Pro | 52.4 | — |
multimodal
| Benchmark | GPT-5.4 nano | Phi 4 Reasoning |
|---|---|---|
| MMMU-Pro | 66.1 | — |
agent
| Benchmark | GPT-5.4 nano | Phi 4 Reasoning |
|---|---|---|
| Toolathlon | 35.5 | — |
| MCP Atlas | 56.1 | — |
general
| Benchmark | GPT-5.4 nano | Phi 4 Reasoning |
|---|---|---|
| MRCR v2 | 33.1 | — |
math
| Benchmark | GPT-5.4 nano | Phi 4 Reasoning |
|---|---|---|
| AIME 2025 | — | 62.9 |
Category Winners
coding
20.0
math
22.0
reasoning
57.8
general
33.3
multimodal
44.4
agent
11.4