Grok 4.6

Grok 4.6 is available through AnyFast at POST /v1/chat/completions. Use model ID grok-4.6.

The upstream xAI specification lists a 500,000-token context window, text and image input, and text output.

When to use Grok 4.6

Long-running coding and agentic tasks
Multi-step knowledge work and complex analysis
Visual understanding with text and image inputs

Quick example

curl https://www.anyfast.ai/v1/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "grok-4.6",
    "messages": [
      {
        "role": "user",
        "content": "Review this function for correctness: function median(a){a.sort();return a[a.length/2]}"
      }
    ],
    "reasoning_effort": "high",
    "stream": false
  }'

from openai import OpenAI

client = OpenAI(
    api_key="YOUR_API_KEY",
    base_url="https://www.anyfast.ai/v1",
)

response = client.chat.completions.create(
    model="grok-4.6",
    messages=[
        {
            "role": "user",
            "content": "Review this function for correctness: function median(a){a.sort();return a[a.length/2]}",
        }
    ],
    reasoning_effort="high",
)

print(response.choices[0].message.content)

Image input

Send image URLs through the OpenAI-compatible multimodal message format:

{
  "model": "grok-4.6",
  "messages": [
    {
      "role": "user",
      "content": [
        {
          "type": "image_url",
          "image_url": {
            "url": "https://example.com/architecture-diagram.png",
            "detail": "high"
          }
        },
        {
          "type": "text",
          "text": "Identify the main components and explain the data flow."
        }
      ]
    }
  ],
  "reasoning_effort": "high"
}

Reasoning behavior

Set the top-level reasoning_effort field:

Value	Behavior
`low`	Lower latency for simpler agentic tasks
`medium`	Balanced reasoning for analysis and tool use
`high`	Default; deeper reasoning for difficult tasks
`xhigh`	Maximum reasoning depth and latency

Reasoning cannot be disabled for Grok 4.6. Do not send reasoning_effort: "none". The upstream model also rejects stop, presence_penalty, and frequency_penalty.

Core request fields

Parameter	Type	Required	Description
`model`	string	Yes	Must be `grok-4.6`
`messages`	array	Yes	Conversation messages. User content can be text or an array containing text and image URL parts
`reasoning_effort`	string	No	`low`, `medium`, `high`, or `xhigh`. Default: `high`
`max_tokens`	integer	No	Maximum number of completion tokens
`stream`	boolean	No	Return SSE deltas when `true`. Default: `false`

Response and streaming

For non-streaming requests, read the final answer from choices[0].message.content. For streaming requests, concatenate the text deltas received from the SSE stream. usage.completion_tokens_details.reasoning_tokens reports the reasoning-token count. These tokens are included in usage.total_tokens; the final answer remains in choices[0].message.content.

This page documents the core AnyFast Chat Completions interface. xAI-native Responses API features such as server-side web search, X search, code execution, conversation storage, and encrypted reasoning are not implied by this endpoint.

API Reference

View the Grok 4.6 request and response schema.

See the official Grok 4.6 model page for the upstream model specification.

Getting started

Model Introduction

Resources

When to use Grok 4.6

Quick example

Image input

Reasoning behavior

Core request fields

Response and streaming

API Reference

​When to use Grok 4.6

​Quick example

​Image input

​Reasoning behavior

​Core request fields

​Response and streaming

API Reference

When to use Grok 4.6

Quick example

Image input

Reasoning behavior

Core request fields

Response and streaming