Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | DeepSeek R1 Distill Qwen 14B | GLM-4.5-Air |
|---|---|---|
| GPQA Diamond | 59.1 | 75 |
| Humanity's Last Exam | — | 10.6 |
coding
| Benchmark | DeepSeek R1 Distill Qwen 14B | GLM-4.5-Air |
|---|---|---|
| SWE-bench Verified | — | 57.6 |
| TerminalBench | — | 30 |
| SciCode | — | 37.3 |
agent
| Benchmark | DeepSeek R1 Distill Qwen 14B | GLM-4.5-Air |
|---|---|---|
| BrowseComp | — | 21.3 |
| TAU-Bench Retail | — | 77.9 |
Category Winners
coding
Z
GLM-4.5-Air27.2
reasoning
Z
GLM-4.5-Air39.7
agent
Z
GLM-4.5-Air39.8