Tavily Web API

aid: tavily:web-api
name: Tavily Web API
description: The Tavily Web API offers a unified set of endpoints for AI agents to search the web, extract
  page content, crawl sites, map sitemaps, and run AI research tasks. Endpoints are REST-based and authenticated
  with a bearer API key.
humanURL: https://docs.tavily.com
baseURL: https://api.tavily.com
tags:
- Search
- Extract
- Crawl
- Map
- Research
- REST
properties:
- type: Documentation
  url: https://docs.tavily.com
- type: GettingStarted
  url: https://docs.tavily.com/welcome
- type: SignUp
  url: https://app.tavily.com/home
- type: APIReference
  url: https://docs.tavily.com/api-reference
- type: SDK
  url: https://github.com/tavily-ai/tavily-python
- type: SDK
  url: https://github.com/tavily-ai/tavily-js
- type: GitHubRepository
  url: https://github.com/tavily-ai
- type: Pricing
  url: https://www.tavily.com/pricing
- type: Authentication
  url: https://docs.tavily.com/documentation/api-credentials
features:
- name: Search
  description: Real-time web search with topic, depth, and time-range controls, p50 latency around 180ms.
- name: Extract
  description: Pull clean text and metadata from one or many URLs.
- name: Crawl
  description: Crawl websites with natural-language instructions and constraints.
- name: Map
  description: Generate a structured map of a site's pages and structure.
- name: Research
  description: Run AI-driven research tasks that aggregate, rank, and cite sources.
- name: Built-in Safeguards
  description: Content validation and safety controls for agent-driven web access.
- name: Drop-in Integrations
  description: First-class adapters for OpenAI, Anthropic, and Groq tool use.
- name: Predictable Latency
  description: Engineered for high-throughput agent workloads with a 99.99% uptime SLA.
useCases:
- name: Agent Web Search Tool
  description: Equip agents with a fast, citation-friendly web search tool.
- name: Real-Time Grounding
  description: Ground LLM responses in fresh web context to reduce hallucinations.
- name: Site Extraction Pipelines
  description: Crawl and extract content from competitor sites or documentation.
- name: AI Research Workflows
  description: Run multi-source research tasks for analysts and knowledge workers.
- name: RAG Backends
  description: Use search and extract endpoints as the retrieval layer behind RAG apps.
integrations:
- name: OpenAI
- name: Anthropic
- name: Groq
- name: LangChain
- name: LangGraph
- name: LlamaIndex
- name: CrewAI
- name: AutoGen
- name: Vercel AI SDK
- name: Model Context Protocol
- name: Databricks
- name: IBM
authentication:
- type: API Key
  description: Bearer token authentication using API keys prefixed with tvly- in the Authorization header.