According to Jinshi’s report on August 1, when Tom Cat was surveyed by the organization, it stated that the company’s overseas R&D team tested GPT on the basis of pre-training language models such as GPT-3.5 and GPT-4 in the early stage, combined with technical services such as Embeddings. The innovative mode combined with the preset database aims to improve the user's smart chat experience while reducing costs. At the same time, the company's overseas team is also testing Google's Bard model to continuously optimize and enrich the interactive experience of the product. According to the previous small-scale overseas reverse payment test, the overseas R&D team has upgraded functions such as active chat, content filtering, and random interaction, and further carried out research and development work on product extension game content, UI manifestations, and payment models.