DeepSeek opens Search and Fetch APIs at $0.75 per 1,000 calls

DeepSeek opened two new APIs to all users on October 17, 2026: Search, which returns web results from the company’s own search engine, and Fetch, which returns a single web page as Markdown. Each costs $0.75 per 1,000 calls. The change that matters most for many developers is a billing date: server-side tool calls made through DeepSeek Harness, Claude Code and Codex, which run on these APIs, are free now and start being billed on October 31, 2026 at 16:00 UTC (00:00 Beijing time on November 1).

What it costs

Item	Price
Search API	$0.75 per 1,000 successful requests
Fetch API	$0.75 per 1,000 successful requests
Server-side tool calls (Harness, Claude Code, Codex)	Free until Oct 31, 16:00 UTC, then per request

DeepSeek’s guide says only HTTP 200 responses that return usable results are billed. Empty results, failures and timeouts are free, and the number of results requested does not change the price. Charges come out of the same prepaid balance as model usage, with granted balance used first.

How the APIs work

• Search (POST /v1/tools/search) takes a query of up to 128 characters and returns 1 to 50 results, each with a title, URL, snippet, summary, publish date and a relevance score. It supports operators for site, language and date filters. The limit is 100 requests per second per account.
• Fetch (POST /v1/tools/fetch) takes one HTTP or HTTPS URL and returns its title and content as Markdown. The limit is 25 requests per second, plus a per-domain limit.

The same release moved deepseek-flash to the DeepSeek-V4.1-Flash-1017 model and retired the deepseek-v4-flash and deepseek-v4-flash-vision-exp names; requests to those names now go to the new model at the Flash price.

Who it affects

Anyone pointing Claude Code, Codex or DeepSeek Harness at DeepSeek’s API, where web search and page fetches happen on DeepSeek’s servers. From October 31 those calls add a per-request charge on top of tokens. DeepSeek does not say how many tool calls a typical coding session makes, so check your usage after the switch. To compare token prices across providers, see our AI token cost calculator.