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 API documentation describes Search and Fetch tools that let agents and retrieval-augmented generation applications search the web and retrieve page content. The listed price is $0.75 per 1,000 successful requests, but the provided material does not specify billing exceptions, availability limits or a launch date.
DeepSeek’s API documentation describes Search and Fetch endpoints that give AI agents and retrieval-augmented generation applications access to live web pages, with pricing listed at $0.75 per 1,000 successful requests. The tools expose search results and retrieved page text through JSON APIs, extending DeepSeek’s developer offering beyond model access, although the documentation supplied does not state when the services became available or define all billing conditions.
The Search API accepts a query and returns structured results from what DeepSeek describes as its in-house search engine. Developers can set the number of results, with a default of 10 and an allowed range of one to 50. A request can also include the user’s original question and an optional IP address for localization. The optional need_content field controls whether page content is included; it defaults to false.
The separate Fetch API retrieves a specified web page and returns its title and content in Markdown. Both are raw JSON endpoints at https://api.deepseek.com/v1/tools/search and https://api.deepseek.com/v1/tools/fetch. The documentation says requests use the same bearer-token authentication as DeepSeek’s API and require an API key and JSON content type.
DeepSeek’s guide lists the price as $0.75 per 1,000 successful requests. That wording ties the quoted rate to successful calls, but the supplied material does not define how success is measured, whether search and fetch calls are charged identically, or whether other costs or limits apply. The guide directs developers to separate reference pages for full request and response schemas.
A Web Retrieval Option for AI Apps
Search and fetch tools can supply an AI application with current web material rather than relying solely on information embedded in a model’s training or on documents a developer has already collected. Search can identify candidate pages, while fetch can bring a page’s text into an agent or retrieval workflow. That makes the endpoints relevant to developers building assistants that need to locate and use external sources.
The listed price gives developers a starting point for estimating usage, but it is not enough on its own to calculate a product’s full operating cost. An application may make multiple calls for one user task, and the documentation provided does not explain whether charges vary by endpoint, result count, content retrieval, or failed requests. Developers will need the full pricing and service terms before projecting costs at scale.
The endpoints also make source handling a design issue for customers. Search results can help locate pages, but the documentation excerpt does not describe how freshness, ranking, coverage, or source quality should be interpreted. Developers remain responsible for deciding how to check retrieved information and how to present citations or uncertainty to end users.
As an affiliate, we earn on qualifying purchases.
How the Two Endpoints Work
The two APIs divide web access into discovery and retrieval. Search accepts a query and returns a results array that can include a title, URL, snippet, summary, content, site name, date and other fields. Some of these fields may be null or empty, depending on availability and request settings. A search with no matches can return an empty results array, according to the guide.
Search queries support operators for restricting or excluding sites, matching exact phrases, specifying languages and setting date boundaries. The guide says the query syntax is forwarded intact. Fetch takes a URL directly and returns the page title and content as Markdown. Its documented URL requirements limit the input to HTTP or HTTPS addresses with a host, with no embedded username or password, and a maximum length of 8,192 characters.
The documentation frames the tools for agents and RAG applications, short for retrieval-augmented generation systems that bring external material into a model workflow. It provides examples in command-line and Python formats, as well as links to full API references. The supplied source does not include independent testing, customer adoption figures or comparisons with other search services.
“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.
Pricing and Service Terms Still Open
The supplied documentation does not identify a publication date or launch date, so it is not possible from this material alone to establish when the endpoints or quoted rate took effect. It also does not provide a complete billing policy, including whether unsuccessful calls, empty search results, retries, or requests that return incomplete page content count toward charges.
Other operational details are absent from the source material: rate limits, geographic availability, uptime commitments, data retention, privacy handling, and any restrictions on fetched pages. The guide refers readers to separate API references, but those full references and service terms were not included here. DeepSeek’s description of its search engine and tool purpose is a company statement, not an independently verified assessment of search quality or coverage.
web content fetcher for AI applications
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Check the Full API and Pricing Terms
Developers considering the tools will need to consult DeepSeek’s Search API and Fetch API references, along with current pricing and service terms, before integrating them. Those materials should clarify request schemas, limits, billing rules and availability. The supplied guide gives sample calls and basic input requirements, but not enough detail to resolve those questions.
Until DeepSeek publishes or makes available further specifics, the confirmed development is limited to the documented endpoints, their basic functions and the listed per-request rate. Whether the service will receive additional capabilities, wider availability or revised pricing is not stated in the source material.
As an affiliate, we earn on qualifying purchases.
Key Questions
What are DeepSeek’s Search and Fetch APIs?
Search accepts a query and returns structured web results. Fetch retrieves a specified web page and returns its title and content in Markdown. DeepSeek presents both as tools for agents and retrieval-augmented generation applications.
How much do the requests cost?
The supplied DeepSeek guide lists a rate of $0.75 per 1,000 successful requests. It does not define all billing cases or clarify whether the two endpoints have separate accounting rules.
How do developers authenticate?
The documentation says the APIs use the same bearer authentication as DeepSeek’s API. Requests require an API key in the Authorization header and a JSON content-type header.
Does Search always return page text?
No. Search’s need_content field defaults to false; when it is false, the content field is empty. Developers can request content in the search response, or use Fetch to retrieve a page by URL.
What is not yet clear from the supplied information?
The source does not specify the launch date, rate limits, availability, service guarantees, privacy terms or exactly how DeepSeek counts a successful request for billing. The full API references and current service terms are needed to answer those questions.
Source: hn
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
