Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | DeepSeek-V4-Flash-Max | Ministral 3 (14B Reasoning 2512) |
|---|---|---|
| GPQA Diamond | 88.1 | 71.2 |
| Humanity's Last Exam | 45.1 | — |
coding
| Benchmark | DeepSeek-V4-Flash-Max | Ministral 3 (14B Reasoning 2512) |
|---|---|---|
| SWE-bench Verified | 79 | — |
| SWE-Bench Pro | 52.6 | — |
general
| Benchmark | DeepSeek-V4-Flash-Max | Ministral 3 (14B Reasoning 2512) |
|---|---|---|
| SimpleQA | 34.1 | — |
agent
| Benchmark | DeepSeek-V4-Flash-Max | Ministral 3 (14B Reasoning 2512) |
|---|---|---|
| BrowseComp | 73.2 | — |
| Toolathlon | 47.8 | — |
| MCP Atlas | 69 | — |
math
| Benchmark | DeepSeek-V4-Flash-Max | Ministral 3 (14B Reasoning 2512) |
|---|---|---|
| AIME 2025 | — | 85 |
Category Winners
coding
50.1
math
61.0
reasoning
82.7
general
43.7
agent
53.1