Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | GPT-5.4 nano | IBM Granite 4.2 8B |
|---|---|---|
| GPQA Diamond | 82.8 | 64.1 |
| Humanity's Last Exam | 24.3 | — |
coding
| Benchmark | GPT-5.4 nano | IBM Granite 4.2 8B |
|---|---|---|
| SWE-Bench Pro | 52.4 | 19.1 |
| SWE-bench Verified | — | 47.7 |
| SciCode | — | 36.1 |
multimodal
| Benchmark | GPT-5.4 nano | IBM Granite 4.2 8B |
|---|---|---|
| MMMU-Pro | 66.1 | — |
agent
| Benchmark | GPT-5.4 nano | IBM Granite 4.2 8B |
|---|---|---|
| Toolathlon | 35.5 | — |
| MCP Atlas | 56.1 | — |
general
| Benchmark | GPT-5.4 nano | IBM Granite 4.2 8B |
|---|---|---|
| MRCR v2 | 33.1 | — |
math
| Benchmark | GPT-5.4 nano | IBM Granite 4.2 8B |
|---|---|---|
| AIME 2025 | — | 86.7 |
Category Winners
coding
28.6
math
I
IBM Granite 4.2 8B64.1
reasoning
55.3
general
27.5
multimodal
43.6
agent
11.1