DeepSeek V4.1 Flash 1017

DeepSeek🇨🇳deepseek/deepseek-v4.1-flash-1017

ChatCompare

This is the October 17 snapshot of DeepSeek V4.1 Flash, a multimodal mixture-of-experts model with a roughly 1M-token context window, tool calling, and adjustable reasoning effort. The Flash tier targets fast, low-cost coding and long-running agent tasks, and the dated id pins behavior to this snapshot.