Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | o4-mini | Qwen 2.5 72B |
|---|---|---|
| GPQA Diamond | 81.4 | 49 |
| Humanity's Last Exam | 14.7 | — |
| BigBench-Hard | — | 83.5 |
| ARC-Challenge | — | 95.3 |
coding
| Benchmark | o4-mini | Qwen 2.5 72B |
|---|---|---|
| SWE-bench Verified | 68.1 | — |
| HumanEval | — | 86.6 |
multimodal
| Benchmark | o4-mini | Qwen 2.5 72B |
|---|---|---|
| MMMU | 81.6 | — |
| CharXiv Reasoning | 72 | — |
agent
| Benchmark | o4-mini | Qwen 2.5 72B |
|---|---|---|
| BrowseComp | 51.5 | — |
| TAU-Bench Retail | 71.8 | — |
language
| Benchmark | o4-mini | Qwen 2.5 72B |
|---|---|---|
| HellaSwag | — | 94 |
| WinoGrande | — | 84.8 |
safety
| Benchmark | o4-mini | Qwen 2.5 72B |
|---|---|---|
| TruthfulQA | — | 62.5 |
general
| Benchmark | o4-mini | Qwen 2.5 72B |
|---|---|---|
| IFEval | — | 84.2 |
| Chatbot Arena Elo | — | 1261 |
| MMLU-Pro | — | 71.1 |
Category Winners
coding
59.0
math
78.5
reasoning
47.9
general
63.7
language
65.7
multimodal
66.2
safety
41.8
agent
44.0