Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | DeepSeek-V4.1-Flash | GPT-5.6 Luna |
|---|---|---|
| GPQA Diamond | 90.9 | 92.3 |
| Humanity's Last Exam | 36.8 | — |
coding
| Benchmark | DeepSeek-V4.1-Flash | GPT-5.6 Luna |
|---|---|---|
| SWE-Bench Pro | — | 62.7 |
multimodal
| Benchmark | DeepSeek-V4.1-Flash | GPT-5.6 Luna |
|---|---|---|
| MMMU-Pro | — | 78.4 |
agent
| Benchmark | DeepSeek-V4.1-Flash | GPT-5.6 Luna |
|---|---|---|
| BrowseComp | — | 83.3 |
| Toolathlon | — | 53.4 |
math
| Benchmark | DeepSeek-V4.1-Flash | GPT-5.6 Luna |
|---|---|---|
| FrontierMath | — | 78.6 |
general
| Benchmark | DeepSeek-V4.1-Flash | GPT-5.6 Luna |
|---|---|---|
| MRCR v2 | — | 41.3 |
Category Winners
coding
74.0
math
95.5
reasoning
89.0
general
38.0
multimodal
77.5
agent
66.8