TL;DR
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
DeepSeek’s Search API documentation lists a price of $0.75 per 1,000 successful requests for its Search and Fetch tools. Search returns structured web results, while Fetch retrieves an individual page as Markdown; the documentation does not clarify several billing details, including how partial failures are counted.
DeepSeek’s API documentation lists Search and Fetch tools for giving agents and retrieval-augmented generation (RAG) applications access to live web content, at a stated price of $0.75 per 1,000 successful requests. The tools provide separate functions—finding pages and retrieving page content—through JSON endpoints that use DeepSeek API-key authentication.
The Search API accepts a query and returns a structured results array. Developers can set how many results to request, provide the user’s original question to help interpret the query, and ask for page content to be included. The documentation sets the default result count at 10, with a permitted range of 1 to 50; content is not requested by default.
The Fetch API takes a page URL and returns its title and content in Markdown. Both tools use HTTP POST requests at api.deepseek.com/v1/tools/search and api.deepseek.com/v1/tools/fetch, respectively. Requests require a bearer token containing a DeepSeek API key and a JSON content-type header.
DeepSeek’s guide describes Search as powered by its in-house search engine. Search results may include a title, URL, snippet, summary, content, site name, date and other metadata. The guide says the content field is empty when the request does not ask for content or when content is unavailable. It directs developers to separate API references for full request and response schemas.
A Metered Route to Live Web Data
The listed price gives developers a stated unit cost for adding web search and page retrieval to agent and RAG workflows. Those systems can use search to discover relevant pages and fetch to bring an individual page’s text into a model workflow, rather than relying solely on information already available to the model.
At the documented rate, the charge is tied to successful requests, not simply to the number of results returned. That distinction could matter for applications that issue repeated queries or fetch many pages, but the guide does not spell out exactly how success is defined for billing. Developers should treat the published figure as a listed rate and confirm the accounting rules and any applicable account terms before estimating total costs.
The endpoints also give developers a direct API option without requiring them to assemble search and page extraction from separate services. That may simplify integration, although the documentation supplied here does not provide performance comparisons, coverage measurements, or independent tests of result quality.
As an affiliate, we earn on qualifying purchases.
How Search Differs From Fetch
The two tools address different stages of gathering online information. Search takes a query and produces a list of results, with optional summaries and page content. Fetch takes a specific URL and returns that page’s content. An application could use Search to identify pages and then Fetch to retrieve selected pages, though the documentation does not say that the tools must be used together.
Search requests can use operators such as site: to restrict results to a domain, language filters, date filters, and terms or phrases to exclude. The optional localization field accepts an IP address to localize results. The guide also says unknown request fields are ignored and that a successful search with no matches returns an empty results array.
The stated price is for successful requests; the supplied documentation does not provide a publication date for the pricing information. It also does not establish whether the rate is promotional, whether it varies by region or account, or how it compares with other providers. The announcement is best understood as a documented API offering and price, not as evidence of a particular level of adoption or search quality.
“The Search API and the Fetch API give agents and RAG applications access to live web content.”
— DeepSeek API documentation
As an affiliate, we earn on qualifying purchases.
Billing Rules Still Need Detail
The documentation lists $0.75 per 1,000 successful requests, but the supplied material does not define what counts as successful for billing. It does not say how timeouts, invalid URLs, unavailable pages, empty search results, retries, or partial responses are treated, or whether Search and Fetch share the same rate under all circumstances.
It is also unclear whether the price includes any limits, taxes, account-specific conditions, or changes over time. The guide does not publish service-level commitments, latency figures, search-index coverage, or an independent assessment of relevance and freshness. Those points should not be inferred from the API examples or response fields.
As an affiliate, we earn on qualifying purchases.
Check the API Terms Before Use
Developers considering the tools can consult DeepSeek’s Search API and Fetch API references for complete schemas, then verify current pricing and billing definitions in the applicable account documentation. Testing representative queries and URLs can help establish whether results and extracted page content suit a particular application.
DeepSeek’s supplied guide does not announce a launch date, future rollout milestone, or planned change to the price. Until the company publishes more detail, the key open issue for cost planning is how its billing system counts successful requests, including empty results and retrieval failures.
As an affiliate, we earn on qualifying purchases.
Key Questions
What does DeepSeek charge for Search and Fetch?
The documentation lists $0.75 per 1,000 successful requests. It does not define all billing cases or say whether account-specific terms apply.
What is the difference between the Search API and Fetch API?
Search accepts a query and returns structured web results. Fetch accepts a specific URL and returns that page’s title and content in Markdown.
Do Search requests always return page content?
No. The Search API’s need_content option defaults to false. The documentation says the content field is empty when content is not requested or is unavailable.
Does an empty search count as a successful request?
DeepSeek’s guide says a search with no matches returns an empty results array, but it does not say whether that response is billed as a successful request. The billing definition remains unclear.
When was the pricing announced?
The supplied source is DeepSeek’s API documentation, but it does not give a publication or announcement date for the listed price.
Source: hn
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
