Overview

Brave LLM Context API delivers pre-extracted, relevance-scored web content optimized for grounding LLM responses in real-time search results. Unlike traditional web search APIs that return links and snippets, LLM Context extracts the actual page content—text chunks, tables, code blocks, and structured data so your LLM or AI agent can reason over it directly.

This makes it ideal for AI agents that need web access as a tool, RAG (Retrieval-Augmented Generation) pipelines, and any application that needs to ground LLM output in fresh, verifiable web content with a single API call.

Key Features

API Reference

Use Cases

LLM Context is perfect for:

• AI Agents: Give your agent a web search tool that returns ready-to-use content in a single call
• RAG Pipelines: Ground LLM responses in fresh, relevant web content
• AI Assistants & Chatbots: Provide factual answers backed by real sources
• Question Answering: Retrieve focused context for specific queries
• Fact Checking: Verify claims against current web content
• Content Research: Gather source material on any topic with one API call

Endpoint

GET https://api.search.brave.com/res/v1/llm/context
POST https://api.search.brave.com/res/v1/llm/context

Authentication: Include your API key in the X-Subscription-Token header.

Quick Start

GET Request

curl -X GET "https://api.search.brave.com/res/v1/llm/context?q=tallest+mountains+in+the+world" \
  -H "X-Subscription-Token: <YOUR_API_KEY>"

POST Request

POST accepts the same query parameters as a JSON request body with Content-Type: application/json. This is useful for complex queries or when parameters exceed URL length limits.

curl -s --compressed -X POST "https://api.search.brave.com/res/v1/llm/context" \
  -H "accept: application/json" \
  -H "Accept-Encoding: gzip" \
  -H "X-Subscription-Token: <YOUR_API_KEY>" \
  -H "Content-Type: application/json" \
  -d '{"q": "tallest mountains in the world"}'

Parameters

Query Parameters

Context Size Parameters

Filtering & Local Parameters

Context Size Guidelines

Adjust context parameters based on your task complexity. For agent tool calls, smaller budgets keep responses fast; for deep research, increase them:

Threshold Modes

The context_threshold_mode parameter controls how aggressively the API filters content for relevance:

Local Recall

The enable_local parameter controls whether location-aware recall is used:

For most use cases, you can omit enable_local entirely and let the API auto-detect based on whether you provide location headers. Set it explicitly only when you need to override the default behavior.

Location-Aware Queries

For local queries (restaurants, businesses, directions), provide location context via headers. Local recall will be enabled automatically when location headers are present, or you can set enable_local=true explicitly.

Not all headers are required. If you have coordinates (X-Loc-Lat and X-Loc-Long), those alone are usually sufficient. Otherwise, provide whichever place-name headers you have available (city, state, country, etc.).

Example using coordinates:

curl -X GET "https://api.search.brave.com/res/v1/llm/context" \
  -H "X-Subscription-Token: <YOUR_API_KEY>" \
  -H "X-Loc-Lat: 37.7749" \
  -H "X-Loc-Long: -122.4194" \
  -G \
  --data-urlencode "q=best coffee shops near me"

Example using place name:

curl -X GET "https://api.search.brave.com/res/v1/llm/context" \
  -H "X-Subscription-Token: <YOUR_API_KEY>" \
  -H "X-Loc-City: San Francisco" \
  -H "X-Loc-State: CA" \
  -H "X-Loc-Country: US" \
  -G \
  --data-urlencode "q=best coffee shops near me"

Example with explicit enable_local=true (no location headers needed):

curl -X GET "https://api.search.brave.com/res/v1/llm/context" \
  -H "X-Subscription-Token: <YOUR_API_KEY>" \
  -G \
  --data-urlencode "q=best coffee shops in san francisco" \
  --data-urlencode "enable_local=true"

When local recall is active, the response may include poi (point of interest) and map fields in addition to the standard generic array.

Goggles (Custom Source Ranking)

Goggles let you tailor which sources ground your LLM to better match your use case. You can restrict results to trusted domains, exclude user-generated content, or boost authoritative sources.

Goggles can be provided as a URL pointing to a hosted goggle file, or as an inline definition passed directly in the goggles parameter. For example, to restrict results to specific documentation sites, use an inline goggle with site rules.

For detailed syntax and examples, see the Goggles documentation.

Response Format

The response contains extracted web content organized into grounding data (by content type) and sources metadata.

Standard Response

{
  "grounding": {
    "generic": [
      {
        "url": "https://example.com/page",
        "title": "Page Title",
        "snippets": [
          "Relevant text chunk extracted from the page...",
          "Another relevant passage from the same page..."
        ]
      }
    ],
    "map": []
  },
  "sources": {
    "https://example.com/page": {
      "title": "Page Title",
      "hostname": "example.com",
      "age": ["Monday, January 15, 2024", "2024-01-15", "380 days ago"]
    }
  }
}

Local Response (with enable_local)

When local recall is active, the response may include POI and map data:

{
  "grounding": {
    "generic": [...],
    "poi": {
      "name": "Business Name",
      "url": "https://business.com",
      "title": "Title of business.com website",
      "snippets": ["Business details and information..."]
    },
    "map": [
      {
        "name": "Place Name",
        "url": "https://place.com",
        "title": "Title of place.com website",
        "snippets": ["Place information and details..."]
      }
    ]
  },
  "sources": {
    "https://business.com": {
      "title": "Business Name",
      "hostname": "business.com",
      "age": null
    }
  }
}

Response Fields

LLM Context vs Answers

Brave offers several complementary approaches for AI-powered search:

When to use LLM Context:

• Giving your AI agent a web search tool it can call autonomously
• Building RAG pipelines with your own LLM
• Need full control over how context is processed and presented
• Want raw extracted content without AI-generated summaries
• Optimizing for speed with single-search retrieval
• Need fine-grained control over token budgets and source filtering

When to use Answers:

• Want end-to-end AI answers with citations
• Need OpenAI SDK compatibility
• Building conversational AI agents or chatbots with built-in search
• Require research mode for thorough, multi-search answers

Learn more about Answers and Summarizer Search.

Best Practices

Token Budget Tuning

• Start with defaults (maximum_number_of_tokens=8192, count=20) for most queries
• Reduce for simple factual lookups to save latency and cost (of your inference)
• Increase for complex research tasks that benefit from more context

Source Quality

• Use Goggles to restrict context to trusted, authoritative sources
• Set context_threshold_mode=strict when precision matters more than recall

Error Handling

• Set a 30-second timeout for requests
• Handle empty grounding.generic arrays gracefully—this means no relevant content was found
• Implement retry logic with exponential backoff for transient failures
• Check rate limit headers and respect the 1-second sliding window

Performance

• Use the smallest count and maximum_number_of_tokens values that meet your needs
• For local queries, provide as many location headers as possible for better results

Changelog

This changelog outlines all significant changes to the Brave LLM Context API in chronological order.

2026-02-06

• Launch Brave LLM Context API at /v1/llm/context
• Support for both GET and POST methods
• Single-search context retrieval with configurable token budgets
• Support for Goggles, local/POI queries, and relevance threshold modes