According to two sources cited by Reuters, DeepSeek has plans to release a new artificial intelligence model, R2, in May. However, they have since redirected their focus to roll out the next model after R1 at the fastest pace possible.
Preliminary information indicates that R2 will have improved abilities in coding and reasoning in English language. DeepSeek, a subsidiary of High-Flyer, an investment company focused on quantitative data or Quant Fund, founded by CEO Liang Wenfeng, who currently holds assets worth tens of billions of yuan. This positioning gives DeepSeek a research and development model for new AI models rather than solely operating as a private company looking to generate revenue. This belief in differentiation has led DeepSeek to develop models distinct from their competitors in the Chinese tech industry. Additionally, being a Quant Fund company, they have previously invested heavily in purchasing GPUs for processing before NVIDIA was banned from exporting its latest chip models by the United States.
DeepSeek has yet to comment on this news report.
Source: Reuters
TLDR: DeepSeek plans to release new AI model R2 but is now focusing on developing the next model after R1 quickly. The company’s background as a Quant Fund has influenced its unique approach to AI model development.
Leave a Comment