Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | Claude Sonnet 4.5 | GLM-5.2 |
|---|---|---|
| GPQA Diamond | 83.4 | 91.2 |
| Humanity's Last Exam | — | 54.7 |
math
| Benchmark | Claude Sonnet 4.5 | GLM-5.2 |
|---|---|---|
| AIME 2025 | 87 | — |
general
| Benchmark | Claude Sonnet 4.5 | GLM-5.2 |
|---|---|---|
| MMMLU | 89.1 | — |
agent
| Benchmark | Claude Sonnet 4.5 | GLM-5.2 |
|---|---|---|
| OSWorld | 61.4 | — |
| TAU-Bench Retail | 86.2 | — |
| Toolathlon | — | 48.2 |
| MCP Atlas | — | 76.8 |
coding
| Benchmark | Claude Sonnet 4.5 | GLM-5.2 |
|---|---|---|
| TerminalBench | 50 | — |
| SWE-Bench Pro | — | 62.1 |
Category Winners
coding
92.8
math
64.7
reasoning
Z
GLM-5.290.2
general
72.5
agent
84.0