DeepSeek debuts DeepSeek-V4.1-Flash, its smallest model built on its new Causal Encoder-Decoder architecture, with 552B backbone parameters and 1M-token context (Reuters)

DeepSeek has launched DeepSeek-V4.1-Flash, its smallest model to date. This new model is built on a Causal Encoder-Decoder architecture, featuring 552B backbone parameters and a 1M-token context. DeepSeek-V4.1-Flash represents an advancement in the company's artificial intelligence offerings, as reported by Reuters.