Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | Claude Haiku 4.5 | Qwen3.6-27B |
|---|---|---|
| GPQA Diamond | 73 | 87.8 |
| Humanity's Last Exam | — | 24 |
math
| Benchmark | Claude Haiku 4.5 | Qwen3.6-27B |
|---|---|---|
| AIME 2025 | 80.7 | — |
coding
| Benchmark | Claude Haiku 4.5 | Qwen3.6-27B |
|---|---|---|
| SWE-bench Verified | 73.3 | 77.2 |
| TerminalBench | 41 | — |
| SWE-Bench Pro | — | 53.5 |
general
| Benchmark | Claude Haiku 4.5 | Qwen3.6-27B |
|---|---|---|
| MMMLU | 83 | — |
agent
| Benchmark | Claude Haiku 4.5 | Qwen3.6-27B |
|---|---|---|
| OSWorld | 50.7 | — |
multimodal
| Benchmark | Claude Haiku 4.5 | Qwen3.6-27B |
|---|---|---|
| MMMU | — | 82.9 |
| MMMU-Pro | — | 75.8 |
| CharXiv Reasoning | — | 78.4 |
Category Winners
coding
70.5
math
53.0
reasoning
62.4
general
48.8
multimodal
Q
Qwen3.6-27B74.4
agent
58.5