DeepSeek Opens Search and Fetch APIs With Per-Request Pricing for Agents
The AMW Read
The launch repeats the Search and Fetch announcement in prior coverage; its immediate significance is a broader developer offering with explicit retrieval pricing, without evidence of competitive performance.
DeepSeek Opens Search and Fetch APIs With Per-Request Pricing for Agents
DeepSeek has launched Search and Fetch APIs for developers building AI agents and applications that retrieve external information, according to TechNode. Search returns structured web results, while Fetch extracts a webpage's title and content as Markdown. Each service costs RMB 5 per 1,000 successful requests. Failed requests, empty search results and timeouts incur no charge, and developers can use their existing DeepSeek API credentials.
The release extends DeepSeek's developer offering beyond model responses into the retrieval tools that agents use to obtain external information. Providing search and page extraction through existing credentials reduces the number of separate services a developer needs to configure for those functions. In the foundation-model market, this is a move toward supplying more of the surrounding agent workflow. The announcement establishes functionality and pricing, but provides no evidence about retrieval quality, coverage or reliability relative to alternatives.
Builders should distinguish the published API rates from the temporary billing treatment of server-side calls through tools including DeepSeek Harness, Claude Code and Codex. DeepSeek says those search and fetching calls are currently unbilled, with charging scheduled to begin at midnight Beijing time on November 1. Developers evaluating the services should estimate successful Search and Fetch requests separately and test whether the returned results and Markdown meet their application's needs before that billing change.
