Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | Claude Opus 4.1 | Gemini 3 Flash |
|---|---|---|
| GPQA Diamond | 80.9 | 90.4 |
| Humanity's Last Exam | — | 43.5 |
| ARC-AGI v2 | — | 33.6 |
math
| Benchmark | Claude Opus 4.1 | Gemini 3 Flash |
|---|---|---|
| AIME 2025 | 78 | 99.7 |
coding
| Benchmark | Claude Opus 4.1 | Gemini 3 Flash |
|---|---|---|
| SWE-bench Verified | 74.5 | 78 |
| TerminalBench | 43.3 | — |
agent
| Benchmark | Claude Opus 4.1 | Gemini 3 Flash |
|---|---|---|
| TAU-Bench Retail | 82.4 | — |
| Toolathlon | — | 49.4 |
| MCP Atlas | — | 57.4 |
multimodal
| Benchmark | Claude Opus 4.1 | Gemini 3 Flash |
|---|---|---|
| MMMU-Pro | — | 81.2 |
| CharXiv Reasoning | — | 80.3 |
| ScreenSpot Pro | — | 69.1 |
Category Winners
coding
77.7
math
83.7
reasoning
76.2
general
73.8
multimodal
73.7
agent
86.7