趋近智
参数
-
上下文长度
1,048.576K
模态
Multimodal
架构
Dense
许可证
Proprietary
发布日期
25 Sept 2025
训练数据截止日期
-
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
Gemini 2.5 Flash with max thinking mode for balanced performance and efficiency. Good coding (67.50 LiveBench Coding) and mathematics (75.35) capabilities. Features thinking transparency at Flash-level latency. September 2025 version with improved performance. Suitable for applications requiring explainable reasoning without Pro-level computational cost.
Google's advanced multimodal models with native understanding of text, images, audio, and video. Features massive context windows up to 2.1M tokens, max thinking modes for complex reasoning, and optimized variants for different performance/cost tradeoffs. Includes Pro, Flash, and Flash Lite variants with configurable thinking capabilities for transparent reasoning.
排名
#22
| 基准 | 分数 | 排名 |
|---|---|---|
Data Analysis LiveBench Data Analysis | 0.73 | ⭐ 7 |
StackUnseen ProLLM Stack Unseen | 0.74 | 8 |
Coding Aider Coding | 0.55 | 21 |
Mathematics LiveBench Mathematics | 0.75 | 23 |
Reasoning LiveBench Reasoning | 0.51 | 34 |
Agentic Coding LiveBench Agentic | 0.23 | 34 |
Coding LiveBench Coding | 0.68 | 35 |
排名
#22
编程排名
#11