DeepSeek Launches V4.1 Flash Model Today, Reduces KV Cache Needs to 1/4 of Prior Version
According to Guru Club, DeepSeek officially launched its new V4.1 Flash model today (Sept 10). The smallest model in the company's new architecture series features native multimodal vision understanding capabilities. The model significantly reduces KV Cache buffer size, cutting HBM (high bandwidth m
DEEPSEEK-2.21%
GateNews·09-10 06:09
