Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | GPT OSS 120B | MiniMax M2 |
|---|---|---|
| GPQA Diamond | 80.1 | 78 |
| Humanity's Last Exam | 14.9 | 12.5 |
agent
| Benchmark | GPT OSS 120B | MiniMax M2 |
|---|---|---|
| TAU-Bench Retail | 67.8 | — |
| BrowseComp | — | 44 |
math
| Benchmark | GPT OSS 120B | MiniMax M2 |
|---|---|---|
| AIME 2025 | — | 78 |
coding
| Benchmark | GPT OSS 120B | MiniMax M2 |
|---|---|---|
| SWE-bench Verified | — | 69.4 |
| TerminalBench | — | 46.3 |
| SciCode | — | 36 |
Category Winners
coding
55.6
math
47.8
reasoning
45.3
agent
50.1