[Digital Today reporter Jinju Hong (홍진주)] Chinese artificial intelligence (AI) company DeepSeek has launched an API that allows web search and document collection functions to be used in external development environments. It is targeting the developer-focused AI infrastructure market, highlighting a price of $0.75 per 1,000 successful requests.

On Oct. 19 local time, online media outlet GIGAZINE reported that DeepSeek unveiled its Search API and Fetch API on Oct. 17. The two APIs are designed to find needed information on the web and collect document content, then connect it to the AI processing workflow. DeepSeek also launched a new AI model, "DeepSeek-V4.1-Flash-1017", on the same day.

The Search API provides a function to find needed information on the web using DeepSeek's proprietary search engine. It expands the web search function previously applied to DeepSeek's existing AI service so external developers can use it in their own applications or systems.

The Fetch API takes the contents of a given webpage in markdown format when a specific URL is entered. After finding relevant documents with the Search API and then pulling the main text with the Fetch API, the collected information can be used for follow-on AI work such as summarising or analysing it.

The two APIs are priced at $0.75 per 1,000 successful requests. GIGAZINE cited a comparison of major search API prices compiled by Artificial Analysis and reported that DeepSeek's fees are cheaper than competing services. It said actual costs may vary depending on the number of successful requests and how the service is used.

Integration with existing AI development tools is also drawing attention. When DeepSeek's API is used in DeepSeek Harness, Anthropic's Claude Code and OpenAI's Codex, among others, the Search API and Fetch API are being used on the server side.

DeepSeek plans to start charging after providing free access to server-side use of the Search API and Fetch API in these development tools until 1 a.m. Japan time on Nov. 1, 2026. Developers using DeepSeek's API through those tools may need to check call volumes and the cost structure after the free period ends.

In particular, if AI coding tools automatically perform web searches and document collection during their workflows, API calls can accumulate even if users do not directly request a search. As actual costs can vary depending on the development environment, it is important to review usage and billing conditions before the free period ends.

This launch is being interpreted as a move by DeepSeek to supply not only AI models but also basic functions that can be used in external applications and development tools. Web search and document collection are functions needed to generate answers based on up-to-date information or for AI agents to perform multiple tasks. Providing them as APIs can reduce the burden on developers to build those functions in their own systems.

DeepSeek is seeking to expand its developer ecosystem by pushing ahead with both a new model launch and opening up its search infrastructure. Key points to watch will be whether its relatively low API fees can drive broader use and whether it can maintain price competitiveness after the free period ends.