Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | Claude Haiku 4.5 | GPT-4 Turbo |
|---|---|---|
| GPQA Diamond | 73 | 49.3 |
| BigBench-Hard | — | 85.1 |
| ARC-Challenge | — | 96.3 |
coding
| Benchmark | Claude Haiku 4.5 | GPT-4 Turbo |
|---|---|---|
| SWE-bench Verified | 73.3 | — |
| TerminalBench | 41 | — |
| HumanEval | — | 87.6 |
general
| Benchmark | Claude Haiku 4.5 | GPT-4 Turbo |
|---|---|---|
| MMMLU | 83 | — |
| IFEval | — | 82.1 |
| MMLU-Pro | — | 64.1 |
| Chatbot Arena Elo | — | 1256 |
agent
| Benchmark | Claude Haiku 4.5 | GPT-4 Turbo |
|---|---|---|
| OSWorld | 50.7 | — |
language
| Benchmark | Claude Haiku 4.5 | GPT-4 Turbo |
|---|---|---|
| HellaSwag | — | 95.3 |
| WinoGrande | — | 87.5 |
safety
| Benchmark | Claude Haiku 4.5 | GPT-4 Turbo |
|---|---|---|
| TruthfulQA | — | 70.1 |
Category Winners
coding
70.5
math
60.3
reasoning
62.2
general
48.8
language
85.2
safety
84.8
agent
58.5