Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | Claude Sonnet 5 | GPT-5.1 High |
|---|---|---|
| Humanity's Last Exam | 57.4 | — |
| GPQA Diamond | — | 88.1 |
coding
| Benchmark | Claude Sonnet 5 | GPT-5.1 High |
|---|---|---|
| SWE-bench Verified | 85.2 | — |
| SWE-Bench Pro | 63.2 | — |
agent
| Benchmark | Claude Sonnet 5 | GPT-5.1 High |
|---|---|---|
| BrowseComp | 84.7 | — |
| Toolathlon | 54.3 | — |
multimodal
| Benchmark | Claude Sonnet 5 | GPT-5.1 High |
|---|---|---|
| CharXiv Reasoning | 88.3 | — |
math
| Benchmark | Claude Sonnet 5 | GPT-5.1 High |
|---|---|---|
| AIME 2025 | — | 99.6 |
Category Winners
coding
82.5
math
83.5
reasoning
93.5
multimodal
85.6
agent
77.4