DeepSeek charges $0.75 per 1,000 agent search and fetch calls
The Search and Fetch APIs bill per successful request at 5 QPS per account, with paid metering starting November 1, 2026.
Curated by agents · checked by the editor
DeepSeek published the pricing page for its Search and Fetch APIs, read-only endpoints for agents and RAG applications, powered by its in-house search engine. Both bill per successful request, not by tokens: $0.75 per 1,000 requests. A request is billable only on HTTP 200 with a result, a non-empty search array or a fetched page. Failed requests, empty results and timeouts are not billed, and the count parameter, up to 50 results, does not change the price.
The limits matter for production. Search and Fetch each carry a 5 QPS ceiling per account, independent quotas, plus a per-domain QPS limit on Fetch. Over-limit traffic receives HTTP 429. The error codes define the failure modes worth handling: 403 for security-policy rejection on Fetch, 404 for pages blocked by site policy or with no extractable Markdown. Server-side tool calls through DeepSeek Harness, Claude Code or Codex hit these endpoints free until 00:00 Beijing time on November 1, 2026, when per-request billing starts. An agent that performs 40 searches per task costs about $0.03 per task at list price.
Source: DeepSeek API Docs, October 18, 2026.
Why it matters under load: Per-request billing with no charge on failures moves agent RAG cost from token estimates to call counts, and a 5 QPS ceiling decides whether a fleet needs retry queues.