Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | Claude 3.5 Haiku | Nemotron 3 Nano (30B A3B) |
|---|---|---|
| ARC-Challenge | 91.8 | — |
| GPQA Diamond | 41.6 | 75 |
| Humanity's Last Exam | — | 15.5 |
safety
| Benchmark | Claude 3.5 Haiku | Nemotron 3 Nano (30B A3B) |
|---|---|---|
| TruthfulQA | 63.5 | — |
general
| Benchmark | Claude 3.5 Haiku | Nemotron 3 Nano (30B A3B) |
|---|---|---|
| IFEval | 78.4 | — |
| MMLU-Pro | 65 | — |
| Chatbot Arena Elo | 1200 | — |
coding
| Benchmark | Claude 3.5 Haiku | Nemotron 3 Nano (30B A3B) |
|---|---|---|
| HumanEval | 88.1 | — |
| SWE-bench Verified | — | 38.8 |
| TerminalBench | — | 8.5 |
| SciCode | — | 33.3 |
Category Winners
coding
65.6
math
83.3
reasoning
41.2
general
30.1
safety
48.3