AI platforms
博查万象
A world-knowledge search engine built for AI, delivering high-precision Web Sear
What is 博查万象?
Bocha Wanxiang (Bocha AI) is a search engine platform designed specifically for artificial intelligence applications, enabling AI Agents, chatbots, and RAG systems to connect with real-time, high-quality world knowledge.Through its core Web Search API, developers can retrieve clean and accurate results from nearly 100 billion web pages and diverse content sources including news, images, videos, encyclopedias, and academic papers, effectively addressing LLM hallucinations and information latency.
The platform also features the bocha-semantic-reranker model, built on a Transformer architecture.With only 80 million parameters, this model achieves ranking performance comparable to much larger models (280M-560M parameters), supporting hybrid keyword-vector search and secondary reranking.This significantly improves the relevance of search results to user queries while offering faster inference speeds and lower costs.
Bocha Wanxiang offers high integration flexibility, supporting access via official REST APIs, Function Calls in frameworks like LangChain, and plugins for major agent development platforms.Whether used for real-time online queries in AI applications or as a data cold-start source for RAG systems, Bocha provides structured, easy-to-parse response data to help developers build smarter applications.
Key features
- Massive KnowledgIndexes nearly 100 billion web pages, covering diverse ecosystem sources including news, images, videos, encyclopedias, travel, and academic content.
- High-Precision SFeatures the bocha-semantic-reranker model based on Transformer architecture, achieving high-relevance ranking with low parameter count to enhance search accuracy.
- Natural LanguageDesigned for AI, it supports natural language query understanding and returns structured data with summaries and key details, ready for direct LLM processing.
- Hybrid Search TeSupports hybrid modes combining keyword and vector search, integrated with semantic reranking to ensure results are both comprehensive and precise.
- Flexible IntegraOffers standard HTTP APIs, LangChain tool call interfaces, and plugins for major agent development platforms to lower integration barriers.
- Real-Time FreshnSupports time-based filtering (e.g., last year) to ensure AI applications receive timely information, suitable for scenarios like news tracking.
How to use it
- Visit the Bocha AI Open Platform website, register an account, and access the console to obtain your API Key.
- Review the developer documentation to understand the request parameters and response formats for the Web Search API and Semantic Reranker API.
- Choose your integration method: you can send HTTP requests directly via curl or configure the SDK in your Python/Node.js projects.
- If using frameworks like LangChain, bind the Bocha search function to the LLM toolchain via Function Calls.
- In agent development platforms, search for and add the 'Bocha Search' plugin, then configure the API Key to invoke it in your workflows.
- Test search queries and verify that the returned JSON structure (such as webPages, snippets, and summaries) meets your application requirements.
Where it fits
- AI Agent KnowledProvides autonomous agents with real-time external information retrieval capabilities, enabling them to answer questions about latest news, stock prices, or events.
- RAG ApplicationServes as the underlying data engine for Retrieval-Augmented Generation systems, providing accurate, cited reference content for vertical domain Q&A.
- Smart Customer SHelps customer service bots break through training data cutoff limits by querying product updates, policy changes, or specific user order statuses in real time.
- Academic & MarkeQuickly aggregates academic papers, industry reports, and news to assist researchers in literature reviews and market trend analysis.
Pricing and free quota
Pricing and any free quota follow the official site.
Who it is for
- AI Application Developers
- LLM Algorithm Engineers
- Enterprise RAG System Builders
- Agent Platform Operators
FAQ
What data sources does Bocha Wanxiang cover?
It covers nearly 100 billion web pages, including news, images, videos, encyclopedias, flight and hotel information, academic content, and other extensive ecosystem sources.
What are the advantages of the bocha-semantic-reranker model?
Based on the Transformer architecture, this model achieves ranking performance comparable to 280M-560M parameter models with only 80M parameters, offering faster inference and lower costs.
How can I integrate Bocha Search into my AI application?
You can call it directly via the official HTTP API, use Function Calls in frameworks like LangChain, or install the Bocha Search plugin in supported agent development platforms.
Does the search result support structured output?
Yes, the API returns standard JSON format including fields like title, URL, snippet, and publication date, designed specifically for LLM parsing and optimization.