Deepseek 发布 V4.1-Flash，大幅降低 AI 智能体内存需求
原标题：New Deepseek model V4.1-Flash cuts memory needs for AI agents
AI 摘要
Deepseek 发布新模型 V4.1-Flash，通过大幅压缩 KV 缓存降低长文本处理成本：快速 GPU 内存中的缓存仅需前代 V4-Flash 约四分之一，永久卸载部分降至约八分之一，相比 V1 每 token 全局 KV 缓存缩小 437 倍。模型采用输入/输出计算拆分，读取输入时每 token 仅激活 80 亿参数、生成时激活 160 亿，并将主 KV 缓存从 FP8 改为 FP4 存储。该模型以 MIT 许可在 Hugging Face 开源，API 价格与 V4-Flash 相同，在编码基准上接近顶级闭源模型，但在复杂科学任务和图像分析上仍有差距。
正文节选
New Deepseek model V4.1-Flash cuts memory needs for AI agents Key Points - Deepseek has released its new AI model V4.1-Flash. It sharply cuts the operating costs of processing long texts by massively shrinking the buffer that job needs. - The model also halves the compute needed for data input. It does that through a technical split that activates less compute for reading information than for generating text later. - On coding tasks, the freely available model matches top closed models from Open