趋近智
Claude 4 Sonnet
参数
-
上下文长度
200K
模态
Text
架构
Dense
许可证
Proprietary
发布日期
15 Jan 2025
训练数据截止日期
-
技术规格
注意力结构
Multi-Head Attention
隐藏维度大小
-
层数
-
注意力头
-
键值头
-
激活函数
-
归一化
-
位置嵌入
Absolute Position Embedding
Claude 4 Sonnet
Standard Claude 4 Sonnet optimized for balanced performance and efficiency. Excellent coding capabilities (80.74 LiveBench Coding), solid language understanding (71.01), and good web development (Elo 1386). Popular choice for production applications requiring reliable performance without thinking overhead. Cost-effective for general-purpose use.
关于 Claude 4
Anthropic's fourth generation Claude models with advanced reasoning, extended context windows up to 200K tokens, and configurable thinking effort levels. Features improved safety alignment, nuanced understanding, and sophisticated task completion. Includes Opus (most capable), Sonnet (balanced), and Haiku (fast) variants, with thinking modes that enable transparent chain-of-thought reasoning for complex problems.
其他 Claude 4 模型
评估基准
排名
#39
| 基准 | 分数 | 排名 |
|---|---|---|
StackEval ProLLM Stack Eval | 0.98 | 🥇 1 |
Coding LiveBench Coding | 0.81 | 🥈 2 |
QA Assistant ProLLM QA Assistant | 0.96 | 6 |
Web Development WebDev Arena | 1386 | 19 |
Agentic Coding LiveBench Agentic | 0.38 | 21 |
Data Analysis LiveBench Data Analysis | 0.65 | 40 |
Mathematics LiveBench Mathematics | 0.60 | 44 |
Reasoning LiveBench Reasoning | 0.40 | 47 |
Graduate-Level QA GPQA | 0.75 | 49 |
排名
排名
#39
编程排名
#17