← Models

Grok 4.1 Fast

xAI · Grok

Deprecated xAI fast tool-calling route whose reasoning and non-reasoning slugs now redirect to Grok 4.3.

Part of Grok family · Other versions: Grok 4.3 , Grok 4.20 , Grok Code Fast 1 , Grok 4
Type
language
Context
2M tokens
Max Output
2M tokens
Status
deprecated
Input
$0.2/1M tok
Output
$0.5/1M tok
API Access
Yes
License
proprietary
agentic tool-use search long-context reasoning enterprise deprecated
Released November 2025 · Updated May 16, 2026

Overview

Freshness note: Model capabilities, limits, and pricing can change quickly. This profile is a point-in-time snapshot last verified on May 16, 2026.

Grok 4.1 Fast is now a deprecated xAI tool-calling route. xAI’s May 2026 retirement guide says both grok-4-1-fast-reasoning and grok-4-1-fast-non-reasoning retired from the API on May 15, 2026 at 12:00 PM PT and now redirect to Grok 4.3.

This page is retained so teams can identify old integrations and migrate deliberately. xAI routes the reasoning variant to Grok 4.3 with low reasoning effort and the non-reasoning variant to Grok 4.3 with reasoning_effort set to none.

Capabilities

Historically, the official launch and pricing pages highlighted this capability profile:

  • Strong tool-calling performance for server-side agents.
  • Large 2M-token context window for long-horizon workflows.
  • Agentic search and research with web, X, code execution, file search, and MCP-style tool support.
  • Better long-context stability than earlier xAI fast models, with explicit emphasis on multi-turn performance.
  • Practical fit for enterprise tasks where speed, cost, and tool use all matter together.

This is no longer the xAI model line to choose for new production agents. Use this page for redirect tracking and use Grok 4.3 with the right reasoning effort for new work.

Technical Details

xAI’s launch materials described Grok 4.1 Fast as a tool-calling model and emphasized reinforcement learning in simulated tool environments. The retiring public API variants are:

  • Two variants: reasoning and non-reasoning.
  • 2M-token context window.
  • Native fit with xAI’s Agent Tools API, including web search, X search, code execution, collections search, and MCP support.

As with other xAI entries in this repo, maxOutput is stored as the documented context ceiling for comparison convenience. Developers should enforce explicit generation limits and latency budgets in their applications.

Pricing & Access

xAI’s public pricing page may still expose historical variant names, but the retirement guide supersedes them for new planning:

  • Input: $0.20 per 1M tokens
  • Output: $0.50 per 1M tokens

Additional notes:

  • Cached input is discounted.
  • Tool calls are billed separately.
  • Large-context requests are priced at a premium.

Historical access options:

  • xAI API
  • xAI Enterprise API
  • Agent Tools API-based workflows

Best Use Cases

Use this page for:

  • identifying old grok-4-1-fast-* request paths
  • testing Grok 4.3 replacement behavior
  • checking whether non-reasoning callers should set reasoning_effort to none
  • documenting historical price and context assumptions

Do not start new deployments on Grok 4.1 Fast. For new xAI text, tool, and code workloads, use Grok 4.3.

Comparisons

  • Grok 4.3 (xAI): Recommended replacement for both reasoning and non-reasoning 4.1 Fast variants.
  • Grok 4.20 (xAI): 2M-context preview lane, not the default replacement for old fast callers.
  • Claude Sonnet 4.6 (Anthropic): Stronger general production baseline outside the xAI ecosystem.