GPT-4 is a large multimodal model capable of processing both image and text inputs and generating human-like text outputs. It demonstrates human-level performance on various professional and academic benchmarks.
发布日期2023年6月13日
参数规模—
上下文长度8K
许可证Proprietary
知识截止2022年12月31日
Benchmarks
评测成绩
| 评测基准 | 类别 | 分数 | 来源 |
|---|---|---|---|
| AI2 Reasoning Challenge (ARC) | reasoning general | 96.3 | 来源 |
| HellaSwag | reasoning | 95.3 | 来源 |
| Uniform Bar Exam | legal reasoning | 90.0 | 来源 |
| SAT Math | math reasoning | 89.0 | 来源 |
| LSAT | reasoning legal general | 88.0 | 来源 |
| Winogrande | reasoning language | 87.5 | 来源 |
| MMLU | general reasoning language math | 86.4 | 来源 |
| DROP | reasoning math | 80.9 | 来源 |
| MGSM | math reasoning | 74.5 | 来源 |
| HumanEval | reasoning code | 67.0 | 来源 |
| MATH | math reasoning | 42.0 | 来源 |
| GPQA | reasoning general | 35.7 | 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 |
|---|---|---|---|---|---|---|---|---|
| LLM Gateway | $30.00 | $60.00 | 8K | — | — | ✓ | ✗ | ✗ |
| OpenAI | $30.00 | $60.00 | 8K | — | — | ✓ | ✗ | ✗ |
| Azure | $60.00 | $120 | 8K | — | — | ✓ | ✗ | ✗ |
| Azure Cognitive Services | $60.00 | $120 | 8K | — | — | ✓ | ✗ | ✗ |
价格单位：美元/百万 token，数据来自社区整理，仅供参考。