> For clean Markdown of any page, append .md to the page URL. > For a complete documentation index, see https://docs.empiriolabs.ai/llms.txt. > For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.empiriolabs.ai/_mcp/server. # Tavily Search > Web search with crawl, extract, and URL mapping for fast, structured retrieval across pages and domains for downstream pipelines. ![Tavily Search](https://media.empiriolabs.ai/model-logos/tavily-search.png) [Tavily](/providers/tavily) · Research & Search `POST /v1/search` Web search with crawl, extract, and URL mapping for fast, structured retrieval across pages and domains for downstream pipelines. ## At a glance | Field | Value | | ------------------- | -------------------------------- | | Model id | `tavily-search` | | Model release date | - | | Input modalities | Text | | Output modalities | Text | | Context window | - | | Weight precision | - | | Features | web\_search, crawl, extract, map | | Native inference | No | | New | No | | Supported endpoints | `POST /v1/search` | ## Pricing | Charge | Spec | Rate | | ------------------------------ | ------------ | -------- | | Search (Basic/Fast/Ultra-Fast) | per search | \$0.0096 | | Search (Advanced) | per search | \$0.019 | | Search (Advanced + Answer) | per search | \$0.029 | | Extract (Basic) | per 5 URLs | \$0.0096 | | Extract (Advanced) | per 5 URLs | \$0.019 | | Crawl/Map (basic) | per 10 pages | \$0.0096 | ## Example request ```bash curl https://api.empiriolabs.ai/v1/search \ -H 'Authorization: Bearer $EMPIRIOLABS_API_KEY' \ -H 'Content-Type: application/json' \ -d '{"model": "tavily-search", "query": "latest LLM benchmarks 2026"}' ``` ## Parameters | Parameter | Type | Required | Default | Description | | ---------------------------- | ------- | -------- | ------------ | -------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | `query` | string | yes | - | The natural-language search query. | | `operation_mode` | enum | no | `"Search"` | search \| extract \| crawl \| map. Determines which Tavily endpoint runs. · Allowed: `Search`, `Crawl`, `Extract`, `Map` | | `search_depth` | enum | no | `"basic"` | basic = fast and cheap, advanced = deeper crawl per result with full content. · Allowed: `basic`, `advanced` | | `search_chunks_per_source` | number | no | `3` | Number of relevant content chunks to extract per source. · Range: 1 – 3 | | `search_topic` | enum | no | `"general"` | Search topic preset (general, news, finance, etc.). Alias of topic. · Allowed: `general`, `news`, `finance` | | `include_answer` | boolean | no | true | basic \| advanced \| false. Returns a synthesized answer alongside the results. | | `include_raw_content` | boolean | no | false | Include the raw HTML content alongside cleaned text. | | `include_images` | boolean | no | false | Include image URLs from the search results. | | `include_image_descriptions` | boolean | no | false | Include AI-generated descriptions of returned images. | | `include_favicon` | boolean | no | false | Include the favicon URL of each source. | | `auto_parameters` | boolean | no | false | Let Tavily auto-pick search\_depth and other knobs based on the query. | | `max_results` | number | no | `5` | Maximum number of results to return (1-50). · Range: 1 – 20 | | `time_range` | enum | no | - | day \| week \| month \| year. Restrict results to this recency window. · Allowed: `day`, `week`, `month`, `year` | | `start_date` | string | no | - | Only include sources published after this date (YYYY-MM-DD). | | `end_date` | string | no | - | Only include sources published before this date (YYYY-MM-DD). | | `country` | string | no | - | Two-letter ISO 3166-1 code (US, GB, etc.) for biasing results to a country. | | `include_domains` | string | no | - | Comma-separated allowlist of domains. Only results from these domains are returned. | | `exclude_domains` | string | no | - | Comma-separated denylist of domains to filter out. | | `topic` | enum | no | `"general"` | Search topic preset. general (default), news (recent news), or finance (markets / business). Also accepted as search\_topic. · Allowed: `general`, `news`, `finance` | | `days` | number | no | - | Recency window for the news topic, in days back from today. · Range: 1 – 365 | | `crawl_limit` | number | no | `10` | Max number of pages to crawl from the starting URL. · Range: 1 – 500 | | `crawl_max_depth` | number | no | `1` | Maximum link-depth from the starting URL. · Range: 1 – 5 | | `crawl_max_breadth` | number | no | `20` | Maximum number of links to follow per page. · Range: 1 – 50 | | `crawl_extract_depth` | enum | no | `"basic"` | How deeply to extract content from each crawled page. · Allowed: `basic`, `advanced` | | `crawl_format` | enum | no | `"markdown"` | Output format for crawled content (markdown \| text \| html). · Allowed: `markdown`, `text` | | `crawl_instructions` | string | no | - | Natural-language instructions for what to look for during crawl. | | `crawl_chunks_per_source` | number | no | - | Number of content chunks to extract from each crawled page. · Range: 1 – 5 | | `crawl_select_paths` | string | no | - | Comma-separated URL paths to include in the crawl. | | `crawl_exclude_paths` | string | no | - | Comma-separated URL paths to exclude from the crawl. | | `crawl_select_domains` | string | no | - | Comma-separated domains the crawl is restricted to. | | `crawl_exclude_domains` | string | no | - | Comma-separated domains the crawl will skip. | | `crawl_allow_external` | boolean | no | false | Allow the crawler to follow links to external domains. | | `crawl_include_images` | boolean | no | false | Include image URLs in the crawl output. | | `crawl_include_favicon` | boolean | no | false | Include each crawled page's favicon URL. | | `crawl_timeout` | number | no | `60` | Crawl timeout in seconds. · Range: 5 – 300 | | `extract_depth` | enum | no | `"basic"` | basic \| advanced. Advanced extraction reads JavaScript-rendered content. · Allowed: `basic`, `advanced` | | `extract_format` | enum | no | `"markdown"` | Output format for extracted content (markdown \| text \| html). · Allowed: `markdown`, `text` | | `extract_query` | string | no | - | Natural-language query to filter / focus the extracted content. | | `extract_chunks_per_source` | number | no | - | Number of content chunks to extract from each URL. · Range: 1 – 5 | | `extract_include_images` | boolean | no | false | Include image URLs in the extract output. | | `extract_include_favicon` | boolean | no | false | Include the source page's favicon URL. | | `extract_timeout` | number | no | `10` | Extract timeout in seconds. · Range: 1 – 60 | | `map_limit` | number | no | `50` | Max number of URLs to map from the starting site. · Range: 1 – 100 | | `map_depth` | number | no | `1` | Maximum link-depth from the starting URL. · Range: 1 – 5 | | `map_breadth` | number | no | `20` | Maximum number of links to follow per page during mapping. · Range: 1 – 50 | | `map_instructions` | string | no | - | Natural-language instructions for what URLs to find during mapping. | | `map_select_paths` | string | no | - | Comma-separated URL paths the map is restricted to. | | `map_exclude_paths` | string | no | - | Comma-separated URL paths the map will skip. | | `map_select_domains` | string | no | - | Comma-separated domains the map is restricted to. | | `map_exclude_domains` | string | no | - | Comma-separated domains the map will skip. | | `map_allow_external` | boolean | no | false | Allow the mapper to follow links to external domains. | | `map_timeout` | number | no | `60` | Map timeout in seconds. · Range: 5 – 300 | | `disable_formatting` | boolean | no | false | Skip the EmpirioLabs Markdown formatting (Sources / Results / Code blocks added by the renderer). The raw upstream Tavily JSON is returned for power users / agents. | ## Notes **Operations** * Search, Crawl, Extract, Map **Search depths** * basic, advanced, fast, ultra-fast **Topics** * general, news, finance --- *Machine-readable schema:* `GET https://api.empiriolabs.ai/v1/models/tavily-search`.