According to the Guosheng computer team, DeepSeek has distilled six small models open-sourced to the community through the output of DeepSeek-R1. Among them, the 32B and 70B models have achieved the effect of benchmarking OpenAI o1-mini in multiple capabilities, and the pricing of DeepSeek-R1 API services is much lower than that of OpenAI o1. The lower cost allows enterprises to develop AI applications with higher efficiency, which is expected to accelerate the domestic AI application from concept to actual implementation. DeepSeek's open-source distillation of small models beyond OpenAI o1-mini is also expected to accelerate the model's landing on the end side.