News Summary

DeepSeek, a prominent Chinese AI startup founded by Liang Wenfeng, has postponed the release of its new R2 artificial intelligence model due to persistent technical difficulties with Huawei’s Ascend AI chips. Initially anticipated to follow the successful R1 model, the R2 model faced challenges during training with the Huawei technology, leading DeepSeek to opt for Nvidia chips for training and Huawei’s Ascend for inference. This delay comes amid heightened speculation, especially after OpenAI’s recent unveiling of GPT-5. The startup, known for its groundbreaking advancements in AI, has not announced a new launch date for the R2 model.