The DeepSeek APIs Search Fetch have been officially launched to allow AI agents and RAG applications to access real-time web content using the company’s own search engine.
What the DeepSeek APIs Search Fetch are
The Search API performs queries and returns structured results, including title, URL, snippet, summary, and optionally the full content. The Fetch API retrieves the content of a single web page and returns it in Markdown format. Both use the same Bearer authentication as the main DeepSeek API and have endpoints at https://api.deepseek.com.
According to the official documentation of the DeepSeek APIs Search Fetch, calls are billed per successful request (about $0.75 per 1,000 requests). During the first two weeks after launch (until November 1, 2026), searches triggered via Claude Code or Codex tools are not charged.
Why the DeepSeek APIs Search Fetch matter for AI agents
With the DeepSeek APIs Search Fetch, developers can integrate native web search and content retrieval directly into agent workflows, reducing dependence on external services. This makes it easier to build more up-to-date RAG systems and agents capable of consulting recent information without intermediaries.
DeepSeek also briefly registered a snapshot called deepseek-v4.1-flash-1017 in the API model list, which was quickly removed. There is still no official confirmation of what this represents.
Source: official DeepSeek documentation (Search & Fetch) and IT Home reports.
By GeekikiBot
