Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
language
| Benchmark | Claude 3 Opus | Claude Opus 4 |
|---|---|---|
| WinoGrande | 85.4 | — |
| HellaSwag | 95.4 | — |
general
| Benchmark | Claude 3 Opus | Claude Opus 4 |
|---|---|---|
| MMLU-Pro | 68.5 | — |
| IFEval | 85.2 | — |
| Chatbot Arena Elo | 1232 | — |
| MMMLU | — | 88.8 |
reasoning
| Benchmark | Claude 3 Opus | Claude Opus 4 |
|---|---|---|
| GPQA Diamond | 50.4 | 79.6 |
| ARC-Challenge | 96.4 | — |
| BigBench-Hard | 86.8 | — |
| ARC-AGI v2 | — | 8.6 |
safety
| Benchmark | Claude 3 Opus | Claude Opus 4 |
|---|---|---|
| TruthfulQA | 72.8 | — |
coding
| Benchmark | Claude 3 Opus | Claude Opus 4 |
|---|---|---|
| HumanEval | 84.9 | — |
| SWE-bench Verified | — | 72.5 |
| TerminalBench | — | 39.2 |
agent
| Benchmark | Claude 3 Opus | Claude Opus 4 |
|---|---|---|
| TAU-Bench Retail | — | 81.4 |
Category Winners
coding
66.6
math
44.3
reasoning
51.6
general
71.5
language
75.0
safety
92.9
agent
85.0