DeepSeek Launches V4.1 Flash-1017, Reduces Token Consumption Over 50%
According to Odaily, DeepSeek today launched its new large language model version V4.1 Flash-1017. The model reduces Token consumption by over 50% across most scenarios compared to the previous version through optimized compression and adaptive decoding strategies. DeepSeek maintained the original A
DEEPSEEK-2.21%
GateNews·10-18 06:00
