Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
safety
| Benchmark | Llama 3.1 70B | Step 3.7 Flash |
|---|---|---|
| TruthfulQA | 60.1 | — |
language
| Benchmark | Llama 3.1 70B | Step 3.7 Flash |
|---|---|---|
| WinoGrande | 85.3 | — |
| HellaSwag | 94.4 | — |
reasoning
| Benchmark | Llama 3.1 70B | Step 3.7 Flash |
|---|---|---|
| ARC-Challenge | 94.8 | — |
| GPQA Diamond | 46.7 | — |
| BigBench-Hard | 81.2 | — |
general
| Benchmark | Llama 3.1 70B | Step 3.7 Flash |
|---|---|---|
| MMLU-Pro | 66.4 | — |
| Chatbot Arena Elo | 1207 | — |
| IFEval | 83.6 | — |
coding
| Benchmark | Llama 3.1 70B | Step 3.7 Flash |
|---|---|---|
| HumanEval | 80.5 | — |
| SWE-Bench Pro | — | 56.3 |
Category Winners
coding
S
Step 3.7 Flash43.1
math
55.6
reasoning
29.9
general
45.0
language
70.8
safety
27.6