xAI Launch Web Search API-LiveSearch is a real-time dynamic search for data sources such as X Platforms, Internet and News

xAI Launchs a web search for API: LiveSearch, integrated into its Grok AI model, allowing real-time dynamic search of data sources such as X platforms, Internet and news through API, thus achieving more time-bound and accurate answers.

Core function

  • ** Real-time data search**: Grok has access to X posts, web pages and trend news to answer queries based on the latest information.

  • Flexible controls: Developers can set search parameters through API, e.g.: Enables/disables search (search_parameters.mode).

  • Limit the number of search results (max_search_resources).

  • Specifies the time frame, domain name (including X platform) or whether to include references.

Transparent reasoning: In conjunction with DeepSearch functionality, Grok can demonstrate the search and reasoning process and provide a clear source of answers.

Use the way

  • API Access: xAI provides LiveSearch API,** is currently in the free Beta phase (as of 5 June 2025)**. Developers can access the API key via xAI official network (x.ai/api) and start building applications.

  • Python SDK: Supporting integration through Python requires setting of XAI_API_KEY environment variables and installing SDK.

  • GUI tool: Community-developed Python applications (e.g. xAI LiveSearch App) provide graphical interfaces to facilitate testing and interaction.

Apply the scene

  • ** Real-time information query**: e.g. news summaries, social media developments or market trends.

  • Auxiliary research: Rapid aggregation of multi-source data to support academic or business analysis.

  • Developer Innovation: Build AI applications based on real-time data, such as chat robots or information aggregation tools.

Working mechanisms

# The core advantage

  • There is no need for manual user management of search logic or tool chain calls;

  • Whether or not to search, when to search and what to search for within the model;

  • Support customisation of parameters such as data source, date range, area, quantity, etc.

♪ ♪ ♪ ♪ ♪ The way to open it ♪ Add fields to the chat requests: “Search_parameters”: { “mode”: “on” / “auto” / “off” } [] (https://assets-v2.circle.so/roc5lmfiu4eazyhex4o6rll0d6ei) Available data source type (as specified by source field) 示例:允许模型查找新闻+X 平台内容: “sources”: [ Chuckles ] {“type”: “news”} {“type”: “x”} Advanced Parameter Configuration Support

out of behavioral description

  • In streaming responses, the fields return only in the last (chunk);

  • All requests for structural support for Python examples call.

Payload = { “Messages”: [“Role”: “user”, “content”: “What’s trying in tech news today”?] “Search_parameters”: { “Mode”: “auto,” “Sources”: [{“type”: “news”}, “Return_citations”: True, “From_date”: “2025-05-20”, “to_date”: “2025-05-21” ♪ I’m so sorry ♪ ♪ I’m so sorry ♪ “model”: “grok-3-latest” https://docs.x.ai/docs/guides/live-search