MIT Technology Review3 分钟
Now read the rest of The Spark
With a new reasoning model that matches the performance of ChatGPT o1, DeepSeek managed to turn restrictions into innovation.