Failed requests, empty Search results, and timeouts are not billed. A request counts only on HTTP 200 when Search returns a non-empty results array or Fetch returns page content.

What are the Search/Fetch rate limits?

Search: 100 QPS per account. Fetch: 25 QPS per account (plus per-domain limits for Fetch). Limits are independent of model concurrency limits and apply at the account level.

DeepSeek Search and Fetch API

Search: POST https://api.deepseek.com/v1/tools/search with JSON body { "query": "...", "count": 10 } (count 1–50).

Fetch: POST https://api.deepseek.com/v1/tools/fetch with JSON body { "url": "https://..." }.

Each API costs $0.75 per 1,000 successful requests. Only HTTP 200 responses with non-empty Search results or Fetch page content are billed.

Checked on 26 Oct 2026 against Search & Fetch guide.

Official sources

Endpoints

API	Method & path
Search	POST https://api.deepseek.com/v1/tools/search
Fetch	POST https://api.deepseek.com/v1/tools/fetch

Send Authorization: Bearer <key> and Content-Type: application/json. Example Search body:

Search request (minimal)

{
  "query": "DeepSeek V4 release",
  "count": 5
}

Billing

Search & Fetch API

DeepSeek opened Search and Fetch to all API accounts on 17 October 2026 (per the API change log). Billing is per successful request, not by tokens or by how many results Search returns.

POST /v1/tools/search

$0.75 per 1,000 successful requests

Fetch

POST /v1/tools/fetch

$0.75 per 1,000 successful requests

API	Endpoint	Price (per 1,000 successful requests)
Search	POST /v1/tools/search	$0.75 USD
Fetch	POST /v1/tools/fetch	$0.75 USD

Each API is $0.75 per 1,000 successful requests. A request is billable only on HTTP 200 when Search returns a non-empty results array or Fetch returns page content; failed requests, empty search results, and timeouts are not billed. The count field only limits how many results Search returns and does not change the per-request price.

Checked on 18 Oct 2026 against Search & Fetch guide.

Full token list prices for chat models live on /api-pricing#search-fetch-api. Harness users: the 2026-10-17 API change log and dsh 0.2.1-rc.2 release note mention enabling Search/Fetch in Web Search settings; configure models and API keys in the Harness Web UI per the Harness quickstart.

Errors and retries

Search/Fetch use additional HTTP statuses (403, 404, 408, 504, etc.). For 408, 429, or 5xx, wait and retry; for 402, top up before retrying. See the Errors section on the Search & Fetch guide.

FAQ

Server-side tool calls through DeepSeek Harness, Claude Code, or Codex are free until 00:00 Beijing time on 1 November 2026; per-request billing starts then.

The Search & Fetch guide and Models & Pricing page include the same note. Direct API calls are billed at the published per-request rate now.

Failed requests, empty Search results, and timeouts are not billed.

Search: 100 QPS per account. Fetch: 25 QPS per account (plus per-domain limits for Fetch).