2.8 Trillion Not the End: Domestic AI Model Kimi K4 to Grow Massive, GPU Is the Bottleneck
Moonshot AI
Chinese AI startup Moonshot AI is reportedly planning a significant scale-up of its Kimi K4 model, far beyond the 2.8 trillion parameters of its predecessor, with GPUs being the main constraint. The news highlights the growing demand for computational resources in China's AI industry.
According to a report from finance.sina.com.cn, Moonshot AI's next-generation large language model, Kimi K4, is expected to dramatically increase in scale, with parameter count potentially exceeding the 2.8 trillion of its predecessor. The main bottleneck for this expansion is the limited supply of high-end GPUs, which are essential for training such massive models. The report suggests that despite the challenges, Moonshot AI is pushing forward to develop more capable AI, indicating a trend of intensifying competition in China's AI sector.
- Сокращения
- GPU = Graphics Processing Unit — графический процессор
Source: DeepSeek (GNews) —
original