Qwen2.5 72B Instruct vs GPT-4 Turbo
Lado a lado. Cada métrica. Cada benchmark.
| Tipo | Qwen2.5 72B Instruct | GPT-4 Turbo |
|---|---|---|
| Provider |  Alibaba Qwen |  OpenAI |
| pontuação média | 53.2 | 51.0 |
| Preço de entrada | $0.12 | $10.00 |
| Preço de saída | $0.39 | $30.00 |
| Janela de contexto | 33K tokens (~16 books) | 128K tokens (~64 books) |
| Lançado em | 2024-09-19 | 2024-04-09 |
| Código aberto | Open Source | Proprietary |
Pontuações de benchmark
9 benchmarks · Qwen2.5 72B Instruct: 6, GPT-4 Turbo: 3
| Benchmark | Categoria | Qwen2.5 72B Instruct | GPT-4 Turbo |
|---|---|---|---|
| BBH | reasoning | 73.1 | 66.8 |
| CMMLU | knowledge | 85.7 | 71.0 |
| GPQA diamond | knowledge | 32.2 | 7.5 |
| HellaSwag | knowledge | 79.7 | 93.7 |
| MATH level 5 | math | 63.2 | 23.0 |
| MMLU | knowledge | 80.4 | 76.5 |
| OTIS Mock AIME 2024-2025 | math | 8.0 | 1.0 |
| TriviaQA | knowledge | 71.9 | 84.8 |
| Winogrande | knowledge | 64.6 | 75.0 |