⌘K

Grok 4.5

grok-4.5

xAI frontier model for coding, agentic software engineering, workflow tasks, and knowledge work. It supports text and image input, a 500K context window, function calling, structured outputs, and configurable reasoning effort. Use it through Vtrix Chat Completions with model ID `grok-4.5`.

Authentication

authorization string required

All APIs require authentication via Bearer Token.

Get API Key:

Visit API Key Management Page to get your API Key

Usage:

Add to request header:

Authorization: Bearer YOUR_API_KEY

Model notes

Compared with the existing Grok 4 document, Grok 4.5 changes the model ID to grok-4.5, keeps text and image input, uses a 500000 token context window, and exposes configurable reasoning with low, medium, and high. The default reasoning effort is high, and reasoning cannot be disabled.

Compared with grok-4.3, the notable field changes are: context window changes from 1000000 to 500000; reasoning_effort no longer accepts none; default reasoning changes from low to high. xAI also documents Chat Completions and Responses API support, function calling, web search, X search, and code execution.

For reasoning requests, avoid sending presence_penalty, frequency_penalty, or stop, because upstream reasoning models reject those parameters.

Parameters

model string required

Model ID to use for the request.

Value: grok-4.5

messages array required

Array of message objects representing the conversation history.

role string required

Message role.

Options: user, assistant, system, developer

content string | array required

Text string or multimodal content array. Supports text and image input.

reasoning_effort string

Controls how much reasoning the model performs before answering. Reasoning cannot be disabled for this model.

Default: high

Options: low, medium, high

max_tokens integer

Maximum tokens to generate in the completion. The effective limit depends on the selected route and upstream response budget.

temperature number

Sampling temperature to use.

Default: 1.0

Range: 0.0 - 2.0

top_p number

Nucleus sampling parameter.

Default: 1.0

Range: 0.0 - 1.0

stream boolean

Whether to stream response incrementally.

Default: false

tools array

List of tools the model may call.

type string required

Tool type.

Options: function

function object required

Function tool definition.

tool_choice string | object

Controls how the model selects tools.

Options: none, auto, required, or an explicit function selection object

response_format object

Format that the model must output. Used for structured outputs.

type string required

Output format type.

Options: text, json_object, json_schema