DeepSeek: DeepSeek V4.1 Flash

{"about":"DeepSeek V4.1 Flash 是 DeepSeek 用於長上下文、程式設計與多步驟推理工作的模型。其核心採用 Mixture of Experts 架構，主體含 5520 億參數，搭配非對稱因果編碼器與解碼器。處理輸入時啟用 80 億參數，生成答案時啟用 160 億參數。這種安排區分了資料閱讀與結果生成；開發商也特別針對大量輸入資料的工作發展此架構。\n\n### 適合哪些工作\n\n- 分析技術說明、往來訊息或一組文件，根據指定問題整理連貫結論。\n- 解釋程式碼、提出修改、設計測試情境並比較實作方案。\n- 將詳細需求轉成一連串行動，列出各階段的中間結果。\n- 比對多個文字來源的需求、找出矛盾並整理成可用規格。\n\n### 如何組織請求\n\n先說明所需結果，例如「找出差異，並提出統一的需求版本」。接著加入資料、比較標準及期望的答案結構。處理程式碼時，適合提供相關片段、觀察到的行為及必須保留的限制。也可分別要求原因說明、建議修補內容與驗證方式，讓結果更便於後續使用。\n\n模型本身支援推理工作與代理式情境。在一般文字請求中，實用結果是計畫、分析或程式碼；實際執行程式與套用修改，則由使用中的應用程式安排。\n\n### 輸入、結果與長篇資料\n\n在 Ailoot，本模型卡接受文字輸入並傳回文字答案。請求中可放入指示、對話歷史、文件摘錄與原始碼。答案可指定為說明、清單、比較表或可直接使用的文字片段。\n\n標示的上下文視窗最多為 1,048,576 個 token。這是工作上下文的大小，並非字元數，需計入請求本身、附加資料與歷史。面對大量文件，請指定來源優先順序及具體問題，並預先描述輸出形式。這有助於聚焦所需內容，也更容易將結果與原文核對。","alias_of":null,"aliases":[],"author":"DeepSeek","canonical_model":"deepseek/deepseek-v4.1-flash","ctx":null,"formats":["chat","gemini","messages","responses"],"id":"deepseek/deepseek-v4.1-flash","input":["text"],"kind":"text","max_out":null,"methods":[{"cache_read":0.0084,"cache_write":null,"cache_write_1h":null,"cache_write_5m":null,"conditional_meter":null,"ctx":null,"family":"anthropic","fast_multiplier":1,"flex_multiplier":1,"fmt":"messages","http":"POST","img_in":null,"img_out":null,"in":0.28,"input":["text"],"label":"Anthropic Messages","max_out":null,"max_tool_calls":null,"out":0.84,"output":["text"],"path":"/v1/messages","request_model_id":"deepseek/deepseek-v4.1-flash","route_label":"Ultra","status_reads":[],"tiers":[],"tool_prices":null,"unit":"token","units":[]},{"cache_read":0.0084,"cache_write":null,"cache_write_1h":null,"cache_write_5m":null,"conditional_meter":null,"ctx":null,"family":"openai","fast_multiplier":1,"flex_multiplier":1,"fmt":"responses","http":"POST","img_in":null,"img_out":null,"in":0.28,"input":["text"],"label":"OpenAI Responses","max_out":null,"max_tool_calls":null,"out":0.84,"output":["text"],"path":"/v1/responses","request_model_id":"deepseek/deepseek-v4.1-flash","route_label":"Ultra","status_reads":[],"tiers":[],"tool_prices":null,"unit":"token","units":[]},{"cache_read":0.0084,"cache_write":null,"cache_write_1h":null,"cache_write_5m":null,"conditional_meter":null,"ctx":null,"family":"openai","fast_multiplier":1,"flex_multiplier":1,"fmt":"chat","http":"POST","img_in":null,"img_out":null,"in":0.28,"input":["text"],"label":"OpenAI Chat","max_out":null,"max_tool_calls":null,"out":0.84,"output":["text"],"path":"/v1/chat/completions","request_model_id":"deepseek/deepseek-v4.1-flash","route_label":"Ultra","status_reads":[],"tiers":[],"tool_prices":null,"unit":"token","units":[]},{"cache_read":0.0084,"cache_write":null,"cache_write_1h":null,"cache_write_5m":null,"conditional_meter":null,"ctx":null,"family":"gemini","fast_multiplier":1,"flex_multiplier":1,"fmt":"gemini","http":"POST","img_in":null,"img_out":null,"in":0.28,"input":["text"],"label":"Gemini generateContent","max_out":null,"max_tool_calls":null,"out":0.84,"output":["text"],"path":"/v1beta/models/deepseek/deepseek-v4.1-flash:generateContent","request_model_id":"deepseek/deepseek-v4.1-flash","route_label":"Ultra","status_reads":[],"tiers":[],"tool_prices":null,"unit":"token","units":[]}],"model_aliases":[{"canonical_model_id":"deepseek/deepseek-v4.1-flash","request_model_id":"deepseek/deepseek-v4.1-flash","route_label":"Ultra"}],"name":"DeepSeek: DeepSeek V4.1 Flash","price":{"in":0.28,"in_label":null,"out":0.84,"out_label":null,"sort_in":0.28,"sort_out":0.84,"varies":false},"public_id":"deepseek/deepseek-v4.1-flash","short":"DeepSeek V4.1 Flash 可根據傳入的上下文，協助分析大量文字資料、撰寫程式碼並解決多步驟問題。"}