Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | Claude 3.7 Sonnet | Claude Haiku 4.5 |
|---|---|---|
| GPQA Diamond | 84.8 | 73 |
math
| Benchmark | Claude 3.7 Sonnet | Claude Haiku 4.5 |
|---|---|---|
| AIME 2025 | 54.8 | 80.7 |
coding
| Benchmark | Claude 3.7 Sonnet | Claude Haiku 4.5 |
|---|---|---|
| SWE-bench Verified | 70.3 | 73.3 |
| TerminalBench | 35.2 | 41 |
multimodal
| Benchmark | Claude 3.7 Sonnet | Claude Haiku 4.5 |
|---|---|---|
| MMMU | 75 | — |
general
| Benchmark | Claude 3.7 Sonnet | Claude Haiku 4.5 |
|---|---|---|
| MMMLU | 86.1 | 83 |
agent
| Benchmark | Claude 3.7 Sonnet | Claude Haiku 4.5 |
|---|---|---|
| TAU-Bench Retail | 81.2 | — |
| OSWorld | — | 50.7 |
Category Winners
coding
70.1
math
52.8
reasoning
80.0
general
61.6
multimodal
68.8
agent
84.7