The Context API allows you to search and retrieve relevant information from both proprietary datasets and the web. By default, it intelligently chooses the most relevant results for your query from all available sources.

​
Installation

pip install valyu

​
Environment Setup

To use the SDK, you’ll need to set up your API key in your environment:

# Set your API key in your env file
VALYU_API_KEY="your-api-key-here"

​
Basic Usage

from valyu import Valyu

# The SDK will automatically use VALYU_API_KEY from environment
valyu = Valyu()

response = valyu.context(
    query="Explain modern portfolio theory",
    search_type="all",  # Can be "proprietary", "web", or "all"
    data_sources=["valyu/valyu-arxiv", "valyu/valyu-wikipedia"],  # Optional
    max_num_results=10,
    max_price=10,
    similarity_threshold=0.4,
    query_rewrite=True
)

# Access the results
for result in response.results:
    print(f"Title: {result.title}")
    print(f"Content: {result.content}")
    print(f"Source: {result.source}")
    print(f"Relevance Score: {result.relevance_score}")
    print(f"Price: ${result.price}")
    print(f"URL: {result.url}")

​
Input Parameters

ParameterTypeRequiredDescription
querystrYesThe question or topic you want information on
search_typestrYesType of search: “proprietary”, “web”, or “all”
data_sourcesList[str]NoSpecific indices or URLs to search over (e.g., [“valyu/valyu-arxiv”, “https://www.valyu.network/”])
max_num_resultsintNoNumber of results to return after reranking (default: 10)
similarity_thresholdfloatNoMinimum similarity score for results (default: 0.4)
query_rewriteboolNoWhether to rewrite query for better performance (default: True)
max_pricefloatYesMaximum allowed price per thousand queries (CPM)

​
Response Structure

{
    "success": bool,              # Whether the request was successful
    "query": str,                 # The original input query
    "error": str,                 # Optional: describes issues if partial response
    "tx_id": str,                 # Transaction ID for this request (useful for feedback)
    "results": [
        {
            "title": str,         # Title of the content
            "url": str,           # Link to the content
            "content": str,       # Extracted text from the source
            "description": str,   # Brief description of the content
            "source": str,        # Which index the data is from
            "price": float,       # Cost of accessing the content (PCM)
            "length": int,        # Character count of the content
            "relevance_score": float,  # Score between 0-1 indicating relevance
            "source_type": str,   # "proprietary" or "web"
            "data_type": str      # "structured" or "unstructured"
        }
    ],
    "results_by_source": {
        "proprietary": int,       # Number of proprietary results
        "web": int               # Number of web results
    },
    "total_deduction_pcm": float,     # Total cost per 1000 queries
    "total_deduction_dollars": float,  # Total cost in USD
    "total_characters": int           # Total characters retrieved
}

​
Examples