Qwen3.8-Max and Kimi K3 Reach Terabyte Scale: What Models Can Ordinary Computers Run?
Alibaba/Qwen
Moonshot AI
As frontier AI models like Qwen3.8-Max and Kimi K3 grow to terabyte scale, ordinary users face challenges in running them locally. The article explores practical alternatives for consumer hardware.
The article discusses the recent trend of large language models (LLMs) from Alibaba's Qwen and Moonshot AI's Kimi reaching terabyte-scale sizes, exemplified by Qwen3.8-Max and Kimi K3, which makes them impractical for consumer devices due to massive memory requirements. It then examines what models can realistically run on standard personal computers, likely focusing on smaller, more efficient models or quantization techniques. The piece aims to guide users on practical choices for local AI deployment.
- Abbreviations
- LLM = Large Language Model — большая языковая модель
Source: Moonshot Kimi (GNews) —
original