DeepSeek search and fetch – $0.75 per 1000 successful requests

The Search API and the Fetch API provision agents and RAG applications admission to live web content:

• Search API finds web pages for a query and returns organized results. It is powered by DeepSeek's in-house hunt engine.
• Fetch API retrieves a sole web leaf and returns its satisfied as Markdown.

Both are raw JSON APIs. They use the identical transporter authentication as the DeepSeek API.

The basis URL is https://api.deepseek.com:

See the Search API Reference and Fetch API Reference for the complete petition and reply schemas.

Authentication​

Send your API key and the JSON satisfied category in the petition headers:

Search API​

Request​

curl https://api.deepseek.com/v1/tools/search \
 -H "Authorization: Bearer <DeepSeek API Key>" \
 -H "Content-Type: application/json" \
 -d '{
 "query": "DeepSeek V4 release",
 "count": 5,
 "need_content": true
 }'

import requests

response = requests.post(
 "https://api.deepseek.com/v1/tools/search",
 headers={
 "Authorization": "Bearer <DeepSeek API Key>",
 },
 json={
 "query": "DeepSeek V4 release",
 "count": 5,
 "need_content": True,
 },
)
response.raise_for_status()

for outcome in response.json()["results"]:
 print(result["title"], result["url"])
 print(result["summary"])

Request fields:

Unknown sectors in the petition build are ignored.

Query Operators​

The query is forwarded alongside its hunt syntax intact. The examples below merge operators alongside a hunt term; you can merge multiple operators in one query.

For example:

{
 "query": "DeepSeek site:github.com OR site:huggingface.co lang:en after:2025",
 "count": 10
}

Response​

A prosperous petition returns HTTP 200 alongside a results array. A normal hunt alongside no matches returns an bare array.

{
 "results": [
 {
 "rerank_score": 2.4,
 "title": "DeepSeek V4 example released",
 "url": "https://example.com/deepseek-v4",
 "snippet": "DeepSeek V4 is now available, alongside tool calling and long-context support.",
 "summary": "DeepSeek has released the V4 model. The publish adds improved tool calling and long-context assistance for delegate workflows.",
 "content": "DeepSeek V4 is now mostly available...",
 "site_name": "Example News",
 "site_icon": "https://example.com/favicon.ico",
 "date": "2026-06-01T00:00:00+00:00",
 "authority": 3
 }
 ]
}

Each item in results contains:

Fetch API​

Request​

curl https://api.deepseek.com/v1/tools/fetch \
 -H "Authorization: Bearer <DeepSeek API Key>" \
 -H "Content-Type: application/json" \
 -d '{
 "url": "https://example.com/article"
 }'

import requests

response = requests.post(
 "https://api.deepseek.com/v1/tools/fetch",
 headers={
 "Authorization": "Bearer <DeepSeek API Key>",
 },
 json={"url": "https://example.com/article"},
)
response.raise_for_status()

page = response.json()
print(page["title"])
print(page["content"])

Request fields:

Response​

{
 "title": "Example article",
 "content": "# Example article\n\nArticle content."
}

Fetching is topic to the mark site's preparedness and admission policies.

Errors​

Errors prosecute the OpenAI-compatible error format. The param site is continually null.

{
 "error": {
 "message": "'count' must be between 1 and 50",
 "type": "invalid_request_error",
 "param": null,
 "code": "invalid_request_error"
 }
}

Search and Fetch are read-only. For HTTP 408, 429, or 5xx errors, delay and retry. For HTTP 402, top up your document before retrying. Do not retry another 4xx errors without changing the request.

Billing​

Search and Fetch are billed per prosperous request, not by tokens or by the figure of returned results:

A petition is billable lone whenever it returns HTTP 200 and produces a outcome (Search must come back a non-empty results array, and Fetch must come back leaf content). Failed requests, bare hunt results, and timeouts are not billed. figure lone controls how many results are returned and does not power the per-request price.

Note: whenever you use the DeepSeek API through DeepSeek Harness, Claude Code, or Codex, server-side tool calls already hit the Search and Fetch APIs. These calls are currently liberated of charge; per-request billing starts at 00:00 Beijing period on November 1, 2026.

Rate Limits​

For all account, Search and Fetch are constricted independently, and traffic to one does not consume the other's quota:

The QPS (requests per second) limits are calculated at the document level, despite of which API key is used, and are distinct from the example concurrency limits. Requests complete the account-level QPS bounds obtain HTTP 429. In supplement to the account-level limit, Fetch is additionally topic to a per-domain QPS limit, and requests complete that bounds additionally obtain HTTP 429.