Yesterday, Rakuten Group unveiled Japan's most extensive artificial intelligence model—Rakuten AI Version 3.0. Nevertheless, shortly following its release, Japanese netizens swiftly pointed out that the model had been trained utilizing DeepSeek-V3, sourced from DeepSeek. While leveraging open-source models for training and fine-tuning is a widely accepted practice in the AI community, Japanese-speaking netizens found it emotionally unsatisfactory that Rakuten Group, having received computing power fees sponsored by the Japanese government, did not undertake independent development of the model.
