IT之家 news, November 13 — This evening (the 13th), DeepSeek announced the official opening of its full-suite app API: by enabling the corresponding parameters in the existing API,you can obtain conversation features such as deep-thinking search and file input; voice and related underlying capabilities have also been opened as standalone interfaces (including OCR, ASR, TTS, etc.), supporting individual or combined calls.
Developers can easily implement complex conversation features in their own applications,and can even replicate the complete DeepSeek App 1:1。
This update directly integrates the various interactive features of the DeepSeek App into the existing API. After enabling the corresponding parameters, the relevant tool calls and context stitching are automatically completed on the server side.
Currently, the V4.1-Flash model (deepseek-flash) supports the following capabilities under the Messages, Responses, and Chat Completions interfaces:
Web search:in non-thinking mode, after enabling the deepseek_web_search tool, the model can automatically initiate web or image searches based on the question and answer with citations;
Deep-thinking search:in thinking mode, after enabling the deepseek_web_search tool, the model can autonomously search the web and read links during multi-step reasoning;
File input:supports uploading files in formats such as PDF, Office, and images; after plain-text extraction or mixed text-and-image parsing, they are stitched into the conversation context for the model to answer;
Link reading:after the deepseek_read_url tool is enabled, the model can automatically identify links in user messages, fetch web content, and answer in context;
Model safety:after enabling the safety-check parameter, DeepSeek will provide basic risk control, reviewing model inputs and outputs and returning a refusal flag when risk is triggered. If you provide application services externally, please be sure to enable this parameter and bear obligations such as content safety review, large-model registration, and AI labeling.
In addition, to make it easier for developers to integrate their own workflows more flexibly, the above underlying capabilities and voice capabilitiesall support being called individually as standalone interfaces, including: Search, Fetch, OCR, ASR, TTS.
Using these interfaces, developers can implement functionality essentially identical to the DeepSeek App. To demonstrate this effect intuitively, DeepSeek has built and open-sourced a lightweight frontend Demo based on the new API, which supports directly importing DeepSeek conversation history.
Demo experience entry:https://cdn.deepseek.com/chat-demo
Sample code entry:https://github.com/deepseek-ai/chat-demo
According to IT之家, when users invoke the DeepSeek API to use extension capabilities, in addition to Token fees, calling fees for the relevant tools will also be charged.
Text/image search: 5 yuan / 1,000 calls
Web scraping: 5 yuan / 1,000 calls
Text recognition: 1.5 yuan / 1,000 pages
Speech recognition: limited-time 50% off (5折), 0.39 yuan / hour after the discount
Speech synthesis: limited-time 50% off (5折), 1.5 yuan / 10,000 characters after the discount