Gemini 1.5 Pro is a mid-size multimodal model optimized for a wide range of reasoning tasks. It can process large amounts of data at once, including 2 hours of video, 19 hours of audio, codebases with 60,000 lines of code, or 2,000 pages of text.
发布日期2024年5月1日
参数规模—
上下文长度—
许可证Proprietary
知识截止2023年11月1日
Benchmarks
评测成绩
| 评测基准 | 类别 | 分数 | 来源 |
|---|---|---|---|
| XSTest | safety | 98.8 | 来源 |
| HellaSwag | reasoning | 93.3 | 来源 |
| GSM8k | math reasoning | 90.8 | 来源 |
| BIG-Bench Hard | reasoning math language | 89.2 | 来源 |
| MGSM | math reasoning | 87.5 | 来源 |
| MATH | math reasoning | 86.5 | 来源 |
| MMLU | general reasoning language math | 85.9 | 来源 |
| Natural2Code | reasoning general | 85.4 | 来源 |
| HumanEval | reasoning code | 84.1 | 来源 |
| MRCR | long_context reasoning general | 82.6 | 来源 |
| Video-MME | multimodal vision reasoning | 78.6 | 来源 |
| MMLU-Pro | language reasoning math general | 75.8 | 来源 |
| WMT23 | language | 75.1 | 来源 |
| DROP | reasoning math | 74.9 | 来源 |
| MathVista | math vision multimodal | 68.1 | 来源 |
| MMMU | multimodal reasoning general | 65.9 | 来源 |
| FunctionalMATH | math reasoning | 64.6 | 来源 |
| PhysicsFinals | physics math reasoning | 63.9 | 来源 |
| GPQA | reasoning general | 59.1 | 来源 |
| Vibe-Eval | multimodal vision general | 53.9 | 来源 |
| HiddenMath | math reasoning | 52.0 | 来源 |
| AMC_2022_23 | math reasoning | 46.4 | 来源 |
| FLEURS | language speech-to-text | 6.7 | 来源 |
Pricing
API 价格对比
暂无 API 价格。