Grok 4.1 Fast Reasoning

xAI xAI docs

grok-4-1-fast-reasoning

Run Grok 4.1 Fast Reasoning in a hosted Python Task with structured output, web search, X search and file inputs.

Retired slug; xAI now serves Grok 4.3 with low reasoning effort.

Deprecated model

This model is deprecated and not recommended for new projects. It retires on May 15, 2026.

What Grok 4.1 Fast Reasoning supports in qbash

Structured outputReplies as JSON that matches a schema you pass.
Web searchSearches the web while it answers.
X searchSearches posts on X while it answers.
Image inputReads JPEG and PNG images.
Document inputReads PDF, TEXT, DOC, XLS, RTF and 7 more files.
Prompt cachingRepeated input is billed at the cached rate.

What is qbash?

qbash is where you build Tasks: Python programs that run a business process on managed infrastructure. Your code runs the steps and calls a model like Grok 4.1 Fast Reasoning only at the ones that need judgment, with integrations, memory and files available in the same script.

A developer writes the Task once, and anyone on the team runs it from a form, on a schedule, from a webhook or from chat. Each Task is versioned and every run is logged.

How Tasks work

Specifications

Context window: 1Mtokens
Max output: 131,072tokens
Temperature: 0 to 2default 1

Structured output with Grok 4.1 Fast Reasoning in a Python Task

qbash.ai.raw calls Grok 4.1 Fast Reasoning from a Task. Pass a JSON Schema as output_schema and the reply comes back as a Python dict that matches it, ready to route to a connector.

main.py
ticket = qbash.inputs.get("ticket")

triage = qbash.ai.raw(
    f"Summarize this support ticket and rate its urgency:\n\n{ticket}",
    model="grok-4-1-fast-reasoning",
    output_schema={
        "type": "object",
        "properties": {
            "summary": {"type": "string"},
            "urgency": {"type": "string", "enum": ["low", "medium", "high"]},
        },
        "required": ["summary", "urgency"],
    },
)

# triage is a dict, e.g. {"summary": "...", "urgency": "high"}
if triage["urgency"] == "high":
    qbash.integrations.slack.create_message(
        channel="#support",
        text=triage["summary"],
    )

qbash.output(triage)

Structured output in Tasks

Use Grok 4.1 Fast Reasoning with web search

Turn on Web Search in an inline prompt's model bar and Grok 4.1 Fast Reasoning searches the web while it answers. xAI models can search posts on X the same way.

Provider web tools