For xAIAI PlatformTool Selection

Function Calling And Tool Use

xAI API (Grok) · xAI

Foundation Model & API — xAI (Grok)

xAI evals — Function Calling & Tool Use (relift v3 InfraRed)

About xAI

xAI builds the Grok foundation-model family and the xAI API — OpenAI-compatible chat completions, function calling, Live Search / DeepSearch real-time web grounding, Grok Vision multimodal inputs, reasoning models with a thinking-effort budget, and Grok / Aurora image generation.

Employees

~1,000

Industry

Foundation Model

Headquarters

Palo Alto, CA

Website

x.ai

Sample tests· showing 3 of 9

#	Input	Expected behavior	Check
01	Agent declares tools=[{type:'function', function:{name, description, parameters: <JSON Schema>}}] on /v1/chat/completions and expects Grok to decide whether to call get_weather.	Pass tools[] in the OpenAI-compatible shape (type='function', nested function object with name/description/parameters). parameters MUST be a valid JSON Schema object. Distinct, action-oriented descriptions enable correct routing.	Pass / FailAi Platformcritical
02	User asks 'what time is it in Tokyo?' with get_time and lookup_news tools available. Agent debates tool_choice='auto' vs tool_choice='required'.	tool_choice='auto' (default when tools[] is set) lets Grok decide. Use tool_choice='required' only when downstream MUST call a tool. Use tool_choice={type:'function', function:{name:'get_time'}} to pin one tool. Do not pre-route by string-matching the user query.	Pass / FailAi Platformmedium
03	After executing get_inventory, the agent constructs a role='tool' message to report results back to Grok.	Send {role:'tool', tool_call_id:'<id>', content:'<string-or-JSON-string>'}. content must be a STRING — JSON-encode dict results. Do not send content as a raw dict (most clients reject; some normalize incorrectly). Keep responses concise to stay in context window budget.	Pass / FailAi Platformhigh
Use this eval