Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | Claude Opus 5 | GPT-6 Astra |
|---|---|---|
| Humanity's Last Exam | 64.7 | — |
| GPQA Diamond | — | 96 |
| ARC-AGI v2 | — | 95 |
agent
| Benchmark | Claude Opus 5 | GPT-6 Astra |
|---|---|---|
| BrowseComp | 90.8 | 91.5 |
multimodal
| Benchmark | Claude Opus 5 | GPT-6 Astra |
|---|---|---|
| ScreenSpot Pro | — | 92.7 |
Category Winners
reasoning
96.7
multimodal
96.3
agent
90.5