tavily

Tavily AI Search

A search API optimized for AI Agents, returning clean, structured results suitable for direct consumption by LLMs.

Use Cases

• Need high-quality search results for complex tasks such as research and writing
• An upgrade option when ddg-search results are not precise enough
• The search backend in the deep-research workflow

Prerequisites

Need to set the environment variable TAVILY_API_KEY：

1. Visit https://tavily.com/ to sign up (free plan: 1000 times per month)
2. Setup:export TAVILY_API_KEY="tvly-xxxxx"

How to Run

Basic Search

curl -s -X POST "https://api.tavily.com/search" \
  -H "Content-Type: application/json" \
  -d '{
    "api_key": "'$TAVILY_API_KEY'",
    "query": "搜索内容",
    "search_depth": "basic",
    "max_results": 5
  }'

Deep Search

curl -s -X POST "https://api.tavily.com/search" \
  -H "Content-Type: application/json" \
  -d '{
    "api_key": "'$TAVILY_API_KEY'",
    "query": "搜索内容",
    "search_depth": "advanced",
    "max_results": 10,
    "include_raw_content": true
  }'

Python SDK

from tavily import TavilyClient

client = TavilyClient(api_key=os.environ["TAVILY_API_KEY"])

response = client.search(
    query="AI agent 市场分析 2026",
    search_depth="advanced",
    max_results=10,
)

for result in response["results"]:
    print(f"Title: {result['title']}")
    print(f"URL: {result['url']}")
    print(f"Content: {result['content'][:200]}")
    print("---")

Parameter Description

Parameter	Value	Description
search_depth	basic / advanced	basic is fast but shallow, advanced is slow but thorough
max_results	1-20	Number of results to return
include_raw_content	bool	Whether to return the raw content of the web pages
include_domains	list	Limit search to specific domains
exclude_domains	list	Exclude domains

Output Specifications

• Search results are annotated with the source URL and relevance score
• advanced search is used for deep research, basic is used for quick queries
• The free quota is limited; for simple queries, prioritize ddg-search