Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | Claude Haiku 4.5 | Llama 3.3 70B |
|---|---|---|
| GPQA Diamond | 73 | 49.5 |
| BigBench-Hard | — | 82.9 |
| ARC-Challenge | — | 95.2 |
coding
| Benchmark | Claude Haiku 4.5 | Llama 3.3 70B |
|---|---|---|
| SWE-bench Verified | 73.3 | — |
| TerminalBench | 41 | — |
| HumanEval | — | 88.4 |
general
| Benchmark | Claude Haiku 4.5 | Llama 3.3 70B |
|---|---|---|
| MMMLU | 83 | — |
| MMLU-Pro | — | 68.9 |
| Chatbot Arena Elo | — | 1247 |
| IFEval | — | 85.3 |
agent
| Benchmark | Claude Haiku 4.5 | Llama 3.3 70B |
|---|---|---|
| OSWorld | 50.7 | — |
safety
| Benchmark | Claude Haiku 4.5 | Llama 3.3 70B |
|---|---|---|
| TruthfulQA | — | 62.8 |
language
| Benchmark | Claude Haiku 4.5 | Llama 3.3 70B |
|---|---|---|
| WinoGrande | — | 86.1 |
| HellaSwag | — | 94.8 |
Category Winners
coding
70.2
math
71.9
reasoning
59.1
general
60.5
language
77.3
safety
43.7
agent
57.3