Robots Atlas>ROBOTS ATLAS
Grok 4.7 Fast
AI Models›Grok

Grok 4.7 Fast

Grok 4.7 Fast (brak publicznego ID modelu) · Family: Grok
The same model as Grok 4.7 but on faster infrastructure: roughly double the output speed at double the price; available only in Cursor and Grok Build, not on the public API.
✓ Active⏳ Limited accessLLMMultimodalReasoning modelTool-using model📁 Grok
Context window
500k
tokens
Release date
21 September 2026
🏢xAIProducer
Access:HostedDeployment:☁ Cloud

Overview

Grok 4.7 Fast is a serving variant of the Grok 4.7 model, released by xAI alongside it on 21 September 2026. It is not a separate, more capable model — in terms of weights and capability it is identical to Grok 4.7. The difference lies purely in infrastructure: the Fast variant runs on faster backing hardware and generates output at roughly double the speed.

That speed is paid for with double the per-token rates of standard Grok 4.7. Availability is distinctly limited: xAI offers the Fast variant only in Cursor and in Grok Build, and it is not part of Grok Build's free tier. The model is not available on the public xAI API and — unlike grok-4.7 — has no public model identifier.

All other parameters match standard Grok 4.7: a 500,000-token context window, text and image input, knowledge through May 2026 and the same benchmark results. The variant makes sense where latency inside an agentic loop matters — interactive coding, for instance — not where higher answer quality is needed.

Classification
LLMMultimodalReasoning modelTool-using model
Family: Grok
Access & deployment
Hosted
Cloud
Weights: Closed
Key parameters
📏 Context: 500k
✓ Tools
📥 Input: text, image, documents

Technical specification

Context window
500k
tokens
Knowledge cutoff
1 May 2026
Knowledge boundary
Features:✓ Tool use
Modalities
⬇ Input
textimagedocuments
⬆ Output
textcodestructured_data

Capabilities and applications

Native model capabilities
Reasoning
The model's ability to reason logically and solve complex problems.
Category: reasoning
Advanced reasoning
The ability to perform multi-step, structured reasoning: analysing problems, planning steps, and drawing conclusions from hypotheses. Reasoning-first models (e.g. GPT-5.1 Thinking) dedicate a portion of inference to chains of thought before responding.
Category: reasoning
Long context
Support for large context windows — tens to hundreds of thousands (or millions) of input tokens. Enables analysis of entire codebases, long documents, and many parallel conversations without losing earlier information. GPT-5.1 supports 400,000 tokens.
Category: language
Coding
Generating, analysing and modifying code in many programming languages. Covers writing functions, debugging, refactoring, code review, and creating tests. Measured by benchmarks such as HumanEval and SWE-bench.
Category: coding
Agentic coding
Multi-hour, multi-step programming tasks performed autonomously by the model: cloning a repository, running tests, iterating on fixes, integrating with CLI tools. Characteristic of Codex variants (GPT-5.1-Codex-Mini, Codex-Max).
Category: coding
Agentic capability
The model's ability to autonomously plan and execute multi-step tasks by sequentially using tools, maintaining context, and adapting to intermediate results.
Category: planning
Tool use
The model's ability to call external functions, APIs and tools during a conversation: calculator, search engine, code editor, database. The model decides when and how to use a tool and interprets its result.
Category: planning
Real-time inference
The model's ability to generate responses with very low latency (>1000 tokens/sec) on specialized inference hardware (e.g. Cerebras WSE), enabling interactive, turn-by-turn collaboration with a human.
Category: coding
Image understanding
Analysing and interpreting the content of images.
Category: vision
Structured output
Producing data in structured formats such as JSON.
Category: structured_generation

Pricing

Sources and related pages