Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
coding
| Benchmark | GPT-5.2 Codex | GPT-5.4 mini |
|---|---|---|
| SWE-Bench Pro | 56.4 | 54.4 |
reasoning
| Benchmark | GPT-5.2 Codex | GPT-5.4 mini |
|---|---|---|
| GPQA Diamond | — | 88 |
| Humanity's Last Exam | — | 28.2 |
multimodal
| Benchmark | GPT-5.2 Codex | GPT-5.4 mini |
|---|---|---|
| MMMU-Pro | — | 76.6 |
agent
| Benchmark | GPT-5.2 Codex | GPT-5.4 mini |
|---|---|---|
| Toolathlon | — | 42.9 |
| MCP Atlas | — | 57.7 |
general
| Benchmark | GPT-5.2 Codex | GPT-5.4 mini |
|---|---|---|
| MRCR v2 | — | 33.6 |
Category Winners
coding
47.1
reasoning
63.1
general
28.5
multimodal
73.5
agent
17.2