# Tavily

Web search API for AI agents that returns ranked results with snippets or full page content, plus extract, crawl and research endpoints.

- Provider: Tavily (https://tavily.com)
- Category: Search the web
- Type: Hosted API
- Pricing model: Credits; $0.008 per credit pay as you go, plans down to $0.005 [1]
- Free tier: 1,000 credits a month, no card [1]
- Authentication: API key (bearer); OAuth for the MCP server; keyless for search and extract [5]
- Connect via: REST API, remote MCP server, Python and JavaScript SDKs, CLI [7]
- Read from the provider's pages on 2026-11-04
- Page: https://supermarket.dev/tools/search-the-web/tavily

## Pricing

### Billing
- Pricing model: Credit-based: free monthly credits, pay as you go, or monthly plans [1]
- Free tier: 1,000 credits a month (1,000 basic searches) [1]
- Card required for free tier: No [1]
- Minimum spend: None for free or pay as you go; the cheapest monthly plan is $30 [2]
- Billing period: Monthly; credits reset on the first day of each month [2]

### Prices
1 credit per basic search, 2 per advanced search. $ per 1,000 requests = credits x 1,000 x price per credit: pay as you go 1 x 1,000 x $0.008 = $8; Growth 1 x 1,000 x $0.005 = $5. Advanced is double.
| Search depth | Credits per request | $ per 1,000 requests, pay as you go | $ per 1,000 requests, Growth plan |
|---|---|---|---|
| `basic`, `fast`, `ultra-fast` | 1 | $8.00 | $5.00 |
| `advanced` | 2 | $16.00 | $10.00 |
Source [4]

### Plans
| Plan | Credits per month | Monthly price | Price per credit |
|---|---|---|---|
| Researcher | 1,000 | Free | - |
| Project | 4,000 | $30 | $0.0075 |
| Bootstrap | 15,000 | $100 | $0.0067 |
| Startup | 38,000 | $220 | $0.0058 |
| Growth | 100,000 | $500 | $0.005 |
| Pay as you go | Per usage | $0.008 per credit | $0.008 |
| Enterprise | Custom | Custom | Custom |
Source [1]

### Other charges
- Extract: 1 credit per 5 successful URLs (basic), 2 per 5 (advanced) [1]
- Research: 4 to 110 credits per request (`mini`); 15 to 250 (`pro`) [1]

## Tools

### MCP server
- Official MCP server: Yes, remote [5]
- Address or package: `https://mcp.tavily.com/mcp/`; local bridge via `npx -y mcp-remote https://mcp.tavily.com/mcp/` [5]
- tools/list without credentials: HTTP 401 with `www-authenticate: Bearer`; tool list not read [14]

### MCP tools
Descriptions are from the docs page; tools/list needs credentials (401 without them), so annotations were not read.
| Tool | What it does | Read-only |
|---|---|---|
| `tavily_search` | Search the web for current, relevant sources. | Not stated |
| `tavily_extract` | Pull clean content from one or more URLs. | Not stated |
| `tavily_crawl` | Read many pages across a site. | Not stated |
| `tavily_map` | Discover a site's structure. | Not stated |
| `tavily_research` | Run multi-source research and return a detailed answer. | Not stated |
Source [5]

### API
| Endpoint | What it does |
|---|---|
| `POST /search` | Web search returning ranked results |
| `POST /extract` | Content from given URLs |
| `POST /crawl` | Crawl a site and extract pages |
| `POST /map` | List a site's URLs |
| `POST /research` | Start a research task |
| `GET /research/{request_id}` | Get research task status and result |
| `GET /usage` | Account usage |
| `POST /feedback` | Not described on fetched pages |
Source [8]

### Features
- Results per query: 0 to 20; default 10 [4]
- Content returned: Snippets of up to 3 chunks per source; optional raw page content in markdown or text, images, favicons [4]
- Search types: `topic`: `general`, `news`, `finance`; `search_depth`: `ultra-fast`, `fast`, `basic`, `advanced` [4]
- Freshness controls: `time_range` (day, week, month, year), `start_date`, `end_date`, `max_age_hours` for cached copies [4]
- Domain filters: `include_domains` (max 300), `exclude_domains` (max 150), `include_domains_mode` `restrict` or `prefer` [4]
- Storing results: Not published

## Authentication

### Methods
| Method | Used for | How |
|---|---|---|
| API key (bearer) | REST API | `Authorization: Bearer tvly-YOUR_API_KEY` |
| OAuth | Remote MCP server | Browser sign-in to a Tavily account when the client connects |
| API key in MCP URL | Remote MCP server | `https://mcp.tavily.com/mcp/?tavilyApiKey=<your-api-key>` |
| Keyless | Search and extract, rate-limited | Header `X-Tavily-Access-Mode: keyless` |
Source [4]

### Getting access
- Who creates the account: A person signs up at the dashboard; keyless access needs no account [6]
- Where to get a key: `https://app.tavily.com` [7]
- Other requirements: Production keys need a paid plan or pay as you go enabled [3]

## Setup

### Addresses
- REST API: `https://api.tavily.com/` [8]
- MCP server: `https://mcp.tavily.com/mcp/` [5]
- OpenAPI spec: `https://docs.tavily.com/documentation/api-reference/openapi.json` [7]
- Docs for agents: `https://docs.tavily.com/llms.txt` [7]
- Status page: `https://status.tavily.com` [13]

### Install
| What | Command |
|---|---|
| MCP, Claude Code | `claude mcp add tavily --transport http https://mcp.tavily.com/mcp/` |
| MCP, Codex | `codex mcp add tavily --url https://mcp.tavily.com/mcp/` |
| Python SDK | `pip install tavily-python` |
| JavaScript SDK | `npm i @tavily/core` |
| CLI | `curl -fsSL https://cli.tavily.com/install.sh \| bash` or `pip install tavily-cli` |
Source [9]

## Limits

### Rate limits
| Endpoint | Development key (per minute) | Production key (per minute) |
|---|---|---|
| Default (search, extract, map) | 100 | 1,000 |
| `/crawl` | 100 | 100 |
| `/research` (create) | 20 | 20 |
| `/usage` (per 10 minutes) | 10 | 10 |
Source [3]

### Other limits
- Over-limit response: `429 Too Many Requests`, optionally with `Retry-After` [3]
- Flex tier: On non-paying plans, searches with `max_age_hours` can be refused with `434` under heavy load [3]
- Domain list size: `include_domains` up to 300, `exclude_domains` up to 150 [4]
- Keyless limits: Not published

## Privacy

### Your data
- Data location: Not published
- Retention: Kept while the account exists, until a deletion request, or until no longer needed [10]
- Used for training: Query data may be used to improve responses; AI functionality inputs may be used to train models [11]
- Zero data retention: Not published
- Query data shared with third parties: May be shared with third-party search index providers in limited situations [10]

### Compliance
- Certifications: SOC 2 and ISO 27001 [2]
- HIPAA: Not published

## Sources

1. Credits & Pricing: https://docs.tavily.com/documentation/api-credits.md
2. Pricing page: https://www.tavily.com/pricing
3. Rate limits: https://docs.tavily.com/documentation/rate-limits.md
4. Search API reference: https://docs.tavily.com/documentation/api-reference/endpoint/search.md
5. Tavily MCP: https://docs.tavily.com/documentation/mcp.md
6. Keyless access: https://docs.tavily.com/documentation/keyless.md
7. llms.txt: https://docs.tavily.com/llms.txt
8. OpenAPI spec: https://docs.tavily.com/documentation/api-reference/openapi.json
9. Tavily CLI: https://docs.tavily.com/documentation/tavily-cli.md
10. Privacy policy: https://www.tavily.com/privacy
11. Terms: https://www.tavily.com/terms
12. Trust Center: https://trust.tavily.com
13. Status: https://status.tavily.com
14. Checked by supermarket.dev