Rumors of the imminent release of DeepSeek's next-generation large-scale model, DeepSeek-R2, have been officially debunked again. Although previous reports indicated the model would be released between August 15th and 30th, Tencent Technology has learned from sources close to the company that there are no plans for a release this month, disappointing eager users and the market.
In fact, speculation about the R2 model's release date has been circulating since the beginning of this year. In March, there was a prediction that the model would be released on March 17th, but this was also officially denied. To date, DeepSeek has not announced any specific release date or technical details for the R2 model. Sources revealed that CEO Liang Wenfeng remains dissatisfied with the model's performance, and the development team is still optimizing it, but it has not yet reached a level ready for official release.
It is understood that the DeepSeek team accelerated development of the R2 model in June of this year. The release, originally scheduled for May, was delayed for various reasons. The new model is expected to significantly improve code generation quality and enhance reasoning capabilities in non-English languages. As competition for large AI models intensifies, DeepSeek may be facing a trade-off between performance and release time while maintaining its technological leadership. This delay also reflects the complexity of current technological iterations in the AI field and the cautious attitude of companies towards product maturity.
