Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | DeepSeek-V3.1 | o1 |
|---|---|---|
| GPQA Diamond | 74.9 | 78 |
| Humanity's Last Exam | 15.9 | — |
math
| Benchmark | DeepSeek-V3.1 | o1 |
|---|---|---|
| AIME 2025 | 49.8 | — |
| FrontierMath | — | 5.5 |
coding
| Benchmark | DeepSeek-V3.1 | o1 |
|---|---|---|
| SWE-bench Verified | 66 | 41 |
| TerminalBench | 31.3 | — |
agent
| Benchmark | DeepSeek-V3.1 | o1 |
|---|---|---|
| BrowseComp | 30 | — |
| TAU-Bench Retail | — | 70.8 |
multimodal
| Benchmark | DeepSeek-V3.1 | o1 |
|---|---|---|
| MMMU | — | 77.6 |
Category Winners
coding
43.7
math
12.9
reasoning
69.1
general
98.5
multimodal
75.0
agent
59.1