# Grok 4

> xAI's frontier reasoning model, RL-trained for native tool use and real-time X search, with a Grok 4 Heavy multi-agent tier.

- Maker: xAI (Grok)
- Line: Grok (flagship)
- Canonical page: https://ai-tldr.dev/models/grok-4/
- Updated: 2026-06-18

## In short

Grok 4 is xAI's flagship large language model, released July 9, 2025 as the company's first model positioned as a frontier reasoning system. It is the successor to Grok 3 and the foundation of the later Grok 4.1 and Grok 4.3 releases. It has a 256K-token context window and costs $3.00 per million input tokens and $15.00 per million output tokens. You can use it through xAI.

## Facts

| Fact | Value | Source |
| --- | --- | --- |
| Context window | 256K (max output 8K) | [1] |
| Input price | $3.00 / 1M tokens | [2] |
| Output price | $15.00 / 1M tokens | [2] |
| Licence | Proprietary · API only | [1] |
| Released | 9 July 2025 | [3] |
| Knowledge cutoff | November 2024 | [1] |

## Sources

1. [Grok 4 0709 — xAI API docs](https://docs.x.ai/developers/models/grok-4-0709)
2. [Grok 4 — benchmarks, pricing & context (LLM Stats)](https://llm-stats.com/models/grok-4)
3. [Grok 4 — official announcement (xAI)](https://x.ai/news/grok-4)

## Overview

Grok 4 is xAI's flagship large language model, released July 9, 2025 as the company's first model positioned as a frontier reasoning system. It is the successor to Grok 3 and the foundation of the later Grok 4.1 and Grok 4.3 releases. Grok 4 was trained with roughly 10x more reinforcement-learning compute than Grok 3, using xAI's Colossus cluster of about 200,000 GPUs, and reinforcement learning was applied at pretraining scale to make tool use a first-class capability rather than an add-on.

What sets Grok 4 apart is native, RL-trained tool use: the model reasons while calling a code interpreter and searching the web and X (formerly Twitter) in real time, then folds the results back into its answer. It accepts text and image input and returns text. xAI shipped it in two forms — standard Grok 4 (a single reasoning agent) and Grok 4 Heavy, a multi-agent tier in which several copies of the model work a problem in parallel and reconcile their answers for higher accuracy on the hardest tasks.

At launch Grok 4 posted frontier scores across reasoning, math, and science benchmarks, and Grok 4 Heavy was the first system to clear 50% on Humanity's Last Exam (text-only subset). The standard model is served via the xAI API as grok-4-0709 with a 256K-token context window (128K in the Grok consumer app). It has since been superseded by cheaper, faster Grok 4.1 and 4.3 releases but remains accessible as a legacy model.

## Benchmarks

- Humanity's Last Exam (with tools): 41% ([source](https://www.datacamp.com/blog/grok-4))
- GPQA: 87.5% ([source](https://www.datacamp.com/blog/grok-4))
- AIME 2025: 91.7% ([source](https://www.datacamp.com/blog/grok-4))
- HMMT 2025: 90% ([source](https://www.datacamp.com/blog/grok-4))
- LiveCodeBench (Jan–May): 79% ([source](https://www.datacamp.com/blog/grok-4))
- USAMO 2025: 37.5% ([source](https://www.datacamp.com/blog/grok-4))
- ARC-AGI-2: 15.9% ([source](https://x.com/arcprize/status/1943168950763950555))

## Pricing

- Input: $3.00 / 1M tokens per 1M tokens
- Output: $15.00 / 1M tokens per 1M tokens

xAI list price for the grok-4-0709 model. The standard Grok 4 tier; the Grok 4 Heavy multi-agent tier is offered through the SuperGrok Heavy consumer plan ($300/month).

Source: https://llm-stats.com/models/grok-4

## Strengths

- Frontier reasoning, math, and graduate-level science performance at launch (GPQA 87.5%, AIME 2025 91.7%, HMMT 2025 90.0%)
- Native, reinforcement-learning-trained tool use — code execution plus real-time web and X search woven into its reasoning
- Strong abstract-reasoning generalization: first model to break the single-digit wall on ARC-AGI-2 (15.9%, independently verified by the ARC Prize Foundation)
- Grok 4 Heavy multi-agent tier pushes the hardest benchmarks higher (50.7% on Humanity's Last Exam, 100% on AIME 2025)
- Large 256K-token API context window for long documents and extended agentic sessions
- Vision (image) input alongside text

## Best for

- Hard STEM and competition-math problem solving (AIME/HMMT/USAMO-style reasoning)
- Graduate-level science Q&A and research assistance
- Agentic workflows that need a model to plan, run code, and search the web/X mid-reasoning
- Coding and code review aided by tool use and a large context window
- Long-document analysis using the 256K-token context window
- Research-grade tasks where Grok 4 Heavy's parallel multi-agent accuracy is worth the higher cost

## How to access

- [xAI](https://docs.x.ai/developers/models/grok-4-0709) — model id `grok-4-0709`

## Grok (flagship) versions

- [Grok 4.3](https://ai-tldr.dev/models/grok-4-3/) (30 April 2026)
- [Grok 4.20](https://ai-tldr.dev/models/grok-4-20/) (March 2026)
- [Grok 4.1](https://ai-tldr.dev/models/grok-4-1/) (17 November 2025)
- Grok 4 (9 July 2025) — this page
- [Grok 3](https://ai-tldr.dev/models/grok-3/) (17 February 2025)
- [Grok 2](https://ai-tldr.dev/models/grok-2/) (20 August 2024)
- [Grok 1.5](https://ai-tldr.dev/models/grok-1-5/) (15 May 2024)
- [Grok 1](https://ai-tldr.dev/models/grok-1/) (3 November 2023)

## FAQ

### When was Grok 4 released?

xAI released Grok 4 on July 9, 2025, announcing it in a livestream alongside the Grok 4 Heavy multi-agent tier. The API model id is grok-4-0709.

### What is the difference between Grok 4 and Grok 4 Heavy?

Standard Grok 4 is a single reasoning agent. Grok 4 Heavy runs several copies of the model in parallel on the same problem and reconciles their answers, which raises accuracy on the hardest benchmarks — for example it reached 50.7% on Humanity's Last Exam and 100% on AIME 2025, versus 41.0% and 91.7% for standard Grok 4. Heavy is offered through xAI's SuperGrok Heavy plan.

### How much does Grok 4 cost via the API?

xAI lists the standard Grok 4 model (grok-4-0709) at $3.00 per million input tokens and $15.00 per million output tokens. Newer Grok 4.1 and 4.3 releases are cheaper, so Grok 4 is now a legacy option.

### What is Grok 4's context window and knowledge cutoff?

Grok 4 has a 256K-token context window in the API (128K in the Grok consumer app) and a maximum output of about 8K tokens. xAI's documentation lists a knowledge cutoff of November 2024. It accepts text and image input and returns text.

## Links

- [Grok 4 — official announcement (xAI)](https://x.ai/news/grok-4)
- [Grok 4 0709 — xAI API docs](https://docs.x.ai/developers/models/grok-4-0709)
- [Grok 4 — benchmarks, pricing & context (LLM Stats)](https://llm-stats.com/models/grok-4)
- [Grok 4 — intelligence & price analysis (Artificial Analysis)](https://artificialanalysis.ai/models/grok-4)
- [Grok 4 launch benchmarks explained (DataCamp)](https://www.datacamp.com/blog/grok-4)
- [ARC Prize — Grok 4 sets ARC-AGI-2 SOTA at 15.9%](https://x.com/arcprize/status/1943168950763950555)

Source: AI/TLDR LLM registry, https://ai-tldr.dev/models/grok-4/