Qwen2.5-72B-Instruct is an instruction-tuned 72 billion parameter language model, part of the Qwen2.5 series. It is designed to follow instructions, generate long texts (over 8K tokens), understand structured data (e.g., tables), and generate structured outputs, especially JSON. The model supports multilingual capabilities across over 29 languages.
发布日期2024年9月19日
参数规模72.7B
上下文长度33K
许可证Qwen
知识截止—
Benchmarks
评测成绩
| 评测基准 | 类别 | 分数 | 来源 |
|---|---|---|---|
| GSM8k | math reasoning | 95.8 | 来源 |
| MT-Bench | communication reasoning general roleplay | 93.5 | 来源 |
| MBPP | reasoning general | 88.2 | 来源 |
| MMLU-Redux | language reasoning math general | 86.8 | 来源 |
| HumanEval | reasoning code | 86.6 | 来源 |
| IFEval | general | 84.1 | 来源 |
| MATH | math reasoning | 83.1 | 来源 |
| AlignBench | general language math reasoning roleplay | 81.6 | 来源 |
| Arena Hard | general reasoning creativity | 81.2 | 来源 |
| MultiPL-E | general language | 75.1 | 来源 |
| MMLU-Pro | language reasoning math general | 71.1 | 来源 |
| LiveCodeBench | reasoning general code | 55.5 | 来源 |
| LiveBench | math reasoning general | 52.3 | 来源 |
| GPQA | reasoning general | 49.0 | 来源 |
Pricing
API 价格对比
| 服务商 | 输入价 | 输出价 | 上下文 | 吞吐(tok/s) | 延迟(s) | 函数调用 | 代码执行 | 联网搜索 |
|---|---|---|---|---|---|---|---|---|
| Cortecs | $0.06 | $0.23 | 33K | — | — | ✓ | ✗ | ✗ |
价格单位：美元/百万 token，数据来自社区整理，仅供参考。