模型多源确认
中国模型需一年追赶OpenAI/Anthropic
精选理由
DeepSeek已备齐RSI系统组件，只需更多数据与算力，一年内或能追上OpenAI/Anthropic
DeepSeek等中国顶尖AI实验室已基本具备构建"RSI"系统的所有组件。中国模型需要更多数据、环境和计算资源来缩小与OpenAI/Anthropic的差距。预计大约需要一年时间才能开始缩小这一差距。
原文 · Teortaxes
I think it'll take about a year for Chinese models to build an OpenAI/Anthropic type system and begin narrowing the gap. The best labs (chiefly DeepSeek) have basically all components for "RSI" assembled and only need more dakka. More data, environments, and of course compute.
