Llama 3.1 405B Instruct is a large language model optimized for multilingual dialogue use cases. It outperforms many available open source and closed chat models on common industry benchmarks. The model supports 8 languages and has a 128K token context length.
发布日期2024年7月23日
参数规模405B
上下文长度128K
许可证Llama 3.1 Community License
知识截止—
Benchmarks
评测成绩
| 评测基准 | 类别 | 分数 | 来源 |
|---|---|---|---|
| ARC-C | reasoning general | 96.9 | 来源 |
| GSM8k | math reasoning | 96.8 | 来源 |
| API-Bank | reasoning | 92.0 | 来源 |
| Multilingual MGSM (CoT) | math reasoning | 91.6 | 来源 |
| HumanEval | reasoning code | 89.0 | 来源 |
| IFEval | general | 88.6 | 来源 |
| MBPP EvalPlus | reasoning general | 88.6 | 来源 |
| MMLU (CoT) | language reasoning math general | 88.6 | 来源 |
| BFCL | general reasoning | 88.5 | 来源 |
| MMLU | general reasoning language math | 87.3 | 来源 |
| DROP | reasoning math | 84.8 | 来源 |
| Multipl-E HumanEval | language general | 75.2 | 来源 |
| MATH | math reasoning | 73.8 | 来源 |
| MMLU-Pro | language reasoning math general | 73.3 | 来源 |
| Multipl-E MBPP | general reasoning | 65.7 | 来源 |
| Nexus | general | 58.7 | 来源 |
| GPQA | reasoning general | 50.7 | 来源 |
| Gorilla Benchmark API Bench | reasoning code | 35.3 | 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 | $0.00 | $0.00 | 128K | — | — | ✓ | ✗ | ✗ |
价格单位：美元/百万 token，数据来自社区整理，仅供参考。