Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | Claude Sonnet 4.6 | GPT-5.1 |
|---|---|---|
| GPQA Diamond | 89.9 | 88.1 |
| Humanity's Last Exam | 49 | — |
| ARC-AGI v2 | 58.3 | — |
coding
| Benchmark | Claude Sonnet 4.6 | GPT-5.1 |
|---|---|---|
| SWE-bench Verified | 79.6 | 76.3 |
general
| Benchmark | Claude Sonnet 4.6 | GPT-5.1 |
|---|---|---|
| MMMLU | 89.3 | — |
agent
| Benchmark | Claude Sonnet 4.6 | GPT-5.1 |
|---|---|---|
| OSWorld | 72.5 | — |
| BrowseComp | 74.7 | — |
| MCP Atlas | 61.3 | — |
math
| Benchmark | Claude Sonnet 4.6 | GPT-5.1 |
|---|---|---|
| AIME 2025 | — | 94 |
| FrontierMath | — | 26.7 |
Category Winners
coding
80.4
math
57.0
reasoning
85.8
general
73.2
multimodal
89.6
agent
60.3