xAI News

Grok Code Fast 1

Thu, 28 Aug 2025 00:00:00 +0000

A speedy daily driver

While today's models are undeniably powerful, they often don't feel purpose-built for agentic coding workflows, where loops of reasoning and tool calls can feel frustratingly slow. As heavy users of agentic coding tools, our engineers saw room for a more nimble, responsive solution optimized for our day-to-day tasks.

We built grok-code-fast-1 from scratch, starting with a brand-new model architecture. To lay a robust foundation, we carefully assembled a pre-training corpus rich with programming-related content. For post-training, we curated high-quality datasets that reflect real-world pull requests and coding tasks.

Throughout the training process, we collaborated closely with our launch partners to refine and sharpen the model’s behavior inside their agentic platforms. grok-code-fast-1 has mastered the use of common tools like grep, terminal, and file editing, and thus should feel right at home in your favorite IDE.

We've teamed up with select launch partners to offer grok-code-fast-1 for free for a limited time, including GitHub Copilot, Cursor, Cline, Roo Code, Kilo Code, opencode, and Windsurf.

Try free on

Cursor

Try free on

GitHub Copilot

Try free on

Cline

Blazing fast

Our inference and supercomputing teams developed several innovative techniques to dramatically accelerate our serving speed, creating a uniquely responsive experience where the model will have already called dozens of tools before you even finish reading the first paragraph of the thinking trace. We've also invested in prompt caching optimizations, regularly achieving cache hit rates above 90% when used with our launch partners.

A versatile programmer

grok-code-fast-1 is exceptionally versatile across the full software development stack and is particularly adept at TypeScript, Python, Java, Rust, C++, and Go. It can complete common programming tasks with minimal oversight, ranging from building zero-to-one projects and providing insightful answers to codebase questions to performing surgical bug fixes.

Example 1 of 2Battle Simulator

Danny Limanseta

@dannylimanseta

It's so quick that I actually had to change up how I work in Cursor.

Grok Code Fast is ridiculously fast, it’s probably the fastest model out there right now. It's so quick that I actually had to change up how I work in Cursor.

I threw together this battle simulator prototype in less than a day with Grok Code Fast on Cursor. Because it responds so fast and follows instructions well, I've found I'm better off giving it smaller, focused tasks. That way I can iterate quickly and steer it exactly where I want.

For this battle simulator, my workflow is to have it plan out big features first, then break them down and execute in phases. Works way better than dumping a huge prompt on it all at once. The battle simulator came together so smoothly this way, just rapid iterations in Cursor until I get it to what I want.

An economical choice

We designed grok-code-fast-1 to be widely accessible, priced at:

$0.20 per million input tokens
$1.50 per million output tokens
$0.02 per million cached input tokens

grok-code-fast-1 was crafted to shine in the tasks developers face every day, striking a compelling balance between performance and cost. Its strength lies in delivering strong performance in a economical, compact form factor, making it a versatile choice for tackling common coding tasks quickly and cost-effectively.

Model Performance

Tokens per Second vs Output Price

Tokens per second(TPS)

190

Output price/ per 1M tokens

$18

Methodology

TPS metrics were calculated by directly measuring response generation speed via each model provider's API, considering only the final response tokens.

- Gemini 2.5 Pro, GPT-5, and Claude Sonnet 4: Measured using their respective public APIs.

- Grok Code Fast 1 and Grok 4: Measured using the xAI API.

- Qwen3-Coder: Hosted on DeepInfra at low precision (fp4), which reduces response quality.

We took a holistic approach to evaluating model performance, blending public benchmarks with real-world testing. On the full subset of SWE-Bench-Verified, grok-code-fast-1 scored 70.8% using our own internal harness.

While benchmarks like SWE-Bench provide valuable insights, we've found they don't fully reflect the nuances of real-world software engineering, particularly the end-user experience in agentic coding workflows.

To guide our model training, we pair these benchmarks with routine human assessments, where experienced developers rate the model's end-to-end performance on everyday tasks. We've also built automated evaluations to track key aspects of behavior, helping us balance trade-offs in design.

When developing grok-code-fast-1, we focused on usability and user satisfaction, guided by real-world human evaluations. The result is a model rated by programmers as fast and reliable for everyday coding tasks.

Grok Code for everyone

For a limited time, we’re excited to offer grok-code-fast-1 for free on exclusive launch partners. Here’s what our launch partners had to say about our the model, which was recently released in stealth under the codename sonic.

Free for a limited time

We’re excited to offer Grok Code Fast 1 for free on exclusive launch partners.

"In early testing, Grok Code Fast has shown both its speed and quality in agentic coding tasks. Empowering developers with powerful tools is a core part of our mission at GitHub Copilot, and this is a compelling new option for our developers."

Mario Rodriguez (@mariorod1)

Chief Product Officer, GitHub

Instructions

Try free on

GitHub Copilot

Prompt Engineering Guide

Our team has crafted a Prompt Engineering Guide with tips on how to get the best from grok-code-fast-1.

The model is generally available via the xAI API, priced at $0.20 / 1M input tokens, $1.50 / 1M output tokens, and $0.02 / 1M cached input tokens.

Open

xAI Cloud Console

Read

Prompt Engineering Guide

What to expect in the next few weeks

Last week, we quietly released grok-code-fast-1 under the codename sonic. During this stealth phase, our team carefully monitored community channels and deployed multiple new model checkpoints to address feedback.

As we advance this new model family, we're excited to iterate rapidly on your input. We highly value the developer community's support and encourage you to freely share all feedback, positive and negative.

We'll focus on delivering consistent updates to grok-code-fast-1, with improvements arriving in days rather than weeks. A new variant that supports multimodal inputs, parallel tool calling, and extended context length is already in training.

Read the grok-code-fast-1 model card here. We’re excited to see what you build!

Announcing Grok for Government

Mon, 14 Jul 2025 00:00:00 +0000

Bringing frontier AI to the frontlines

Today, we’re proud to announce Grok For Government – a suite of frontier AI products available to United States Government customers.

xAI’s mission is to create and propagate AI tools to assist humanity in our quest for understanding and knowledge. Supporting the critical missions of the United States Government is a key part of this mission – bringing the best tools and technologies available in the commercial world to our hard-working public servants.

Americans have led the world through all of society’s great technological innovations, and AI will be no exception. xAI is proud to continue this legacy – which is why we are the only company building on this legacy here in the US and turning shovels into tokens.

Under the umbrella of Grok For Government, we will be bringing all of our world-class AI tools to federal, local, state, and national security customers. These customers will be able to use the Grok family of products to accelerate America – from making everyday government services faster and more efficient to using AI to address unsolved problems in fundamental science and technology.

This includes frontier AI like Grok 4, our latest and most advanced model so far, which brings strong reasoning capabilities with extensive pretraining models. Our government partnerships will also bring to bear tools like Deep Search, Tool Use, and more integrations – all of which are industry-leading commercial products.

We’ve been engaging closely with innovators and leaders in the government to make sure that our offerings are able to deliver the capabilities we need. In addition to our commercial offerings, we will be making some unique capabilities available to our government customers, including:

Custom models for national security and critical science applications available to specific customers.
Forward Deployed Engineering and Implementation Support, with USG cleared engineers
Custom AI-powered applications to accelerate use cases in healthcare, fundamental science, and national security, to name a few examples
Models soon available in classified and other restricted environments
- Partnerships with xAI to build custom versions for specific mission sets

We are especially excited to announce two important milestones for our US Government business – a new $200M ceiling contract with the US Department of Defense, alongside our products being available to purchase via the General Services Administration (GSA) schedule. This allows every federal government department, agency, or office, to access xAI's frontier AI products.

America is the world leader in AI, and this is in no small part due to a tradition of innovation and strong investments in engineering and science. We’re excited to contribute back to the country that made xAI uniquely possible here.

Learn more here or reach out to us directly.

We’re also seeking talented mission driven engineers who want to join the cause. If you’re excited by solving hard problems to empower our nation’s hardest workers, we’d love to hear from you. Reach out to us over email or apply online directly here. We’d love to hear from you.

Grok 4

Wed, 09 Jul 2025 00:00:00 +0000

Grok 4 is the most intelligent model in the world. It includes native tool use and real-time search integration, and is available now to SuperGrok and Premium+ subscribers, as well as through the xAI API. We are also introducing a new SuperGrok Heavy tier with access to Grok 4 Heavy - the most powerful version of Grok 4.

Grok 3 Beta — The Age of Reasoning Agents

Wed, 19 Feb 2025 00:00:00 +0000

We are thrilled to unveil an early preview of Grok 3, our most advanced model yet, blending superior reasoning with extensive pretraining knowledge.

xAI raises $6B Series C

Mon, 23 Dec 2024 00:00:00 +0000

We are partnering with A16Z, Blackrock, Fidelity Management & Research Company, Kingdom Holdings, Lightspeed, MGX, Morgan Stanley, OIA, QIA, Sequoia Capital, Valor Equity Partners and Vy Capital, amongst others.

Bringing Grok to Everyone

Thu, 12 Dec 2024 00:00:00 +0000

Grok is now faster, sharper, and has improved multilingual support. It is available to everyone on the 𝕏 platform.

Grok Image Generation Release

Mon, 09 Dec 2024 00:00:00 +0000

We've enhanced Grok's image generation abilities with a new model, code-named Aurora. Aurora is an autoregressive mixture-of-experts network trained to predict the next token from interleaved text and image data. We trained the model on billions of examples from the internet, giving it a deep understanding of the world. As a result, it excels at photorealistic rendering and precisely following text instructions. Beyond text, the model also has native support for multimodal input, allowing it to take inspiration from or directly edit user-provided images.

Grok's new capabilities are now available on the 𝕏 platform in select countries and will roll out to all users within a week.

Lockheed SR-71 Blackbird flying through an abstract sky.

An astronaut standing on the surface of an alien planet, with a spaceship in the background and multiple moons in the sky.

A volcano surrounded by ice.

A superposition of a cat in a hyperbolic time chamber in the style of Van Gogh.

An origami Cybertruck.

Cherry blossoms beneath a sunset sky.

A sketch of a multi-sided 3d geometric shape on paper.

A dog drinking a cup of tea in a library.

A closeup of a guitar player's hand holding a pick.

A comic of a young man standing by the sea, gazing back over his shoulder with a determined expression. In a speech bubble, printing the text, 'Make it happen, yesterday.'

A burger with double meat patty placed on a plate.

A female warrior holding a sword, with intricate armor and a confident expression.

An abstract composition using geometric shapes and vibrant colors, evoking a sense of energy and movement.

Elon Musk as a character in the animated series Rick and Morty.

A serene mountain lake at sunset, with mist rising from the water and the peaks reflected perfectly in the still surface.

A cyberpunk-inspired city at night, with neon lights, flying cars, and towering skyscrapers.

An elderly person, capturing every wrinkle and expression.

A dewdrop on a spider web, with the intricate patterns of the web and the refraction of light.

Image Generation

Grok can now generate high-quality images across several domains where other image generation models often struggle. It can render precise visual details of real-world entities, text, logos, and can create realistic portraits of humans.

Prompt

Cybertruck under an aurora

Grok

Imagen 3

Flux.1 Pro

Ideogram 2.0

Dall-E 3

Image Editing

Our new image generation model can now take images as input, giving users greater creative control and flexibility. We will release this capability to users on the 𝕏 platform soon.

Prompt

Make the cat anime style

Input image

Output image

Looking Forward

At xAI, we are advancing the frontier of multimodal understanding and generation. If this goal inspires you, we invite you to join us on this journey — we are hiring!

API Public Beta

Mon, 04 Nov 2024 00:00:00 +0000

Starting today, developers can build on our Grok foundation models using our newly released API. We will run a public beta program until the end of 2024 during which everyone will get $25 of free API credits per month.

Grok-2 Beta Release

Tue, 13 Aug 2024 00:00:00 +0000

We are excited to release an early preview of Grok-2, a significant step forward from our previous model Grok-1.5, featuring frontier capabilities in chat, coding, and reasoning. At the same time, we are introducing Grok-2 mini, a small but capable sibling of Grok-2. An early version of Grok-2 has been tested on the LMSYS leaderboard under the name "sus-column-r." At the time of this blog post, it is outperforming both Claude 3.5 Sonnet and GPT-4-Turbo.

Grok-2 and Grok-2 mini are currently in beta on 𝕏, and we are also making both models available through our enterprise API later this month.

Grok-2 language model and chat capabilities

We introduced an early version of Grok-2 under the name "sus-column-r" into the LMArena.ai Chatbot Arena, a popular competitive language model benchmark. It outperforms both Claude and GPT-4 on the LMSYS leaderboard in terms of its overall Elo score.

Overall ELO scores on Chatbot Arena

Win Rate of Grok-2 Against Competing Models on Chatbot Arena

Grok's response quality

Internally, we employ a comparable process to evaluate our models. Our AI Tutors engage with our models across a variety of tasks that reflect real-world interactions with Grok. During each interaction, the AI Tutors are presented with two responses generated by Grok. They select the superior response based on specific criteria outlined in our guidelines. We focused on evaluating model capabilities in two key areas: following instructions and providing accurate, factual information. Grok-2 has shown significant improvements in reasoning with retrieved content and in its tool use capabilities, such as correctly identifying missing information, reasoning through sequences of events, and discarding irrelevant posts.

Benchmarks

We evaluated the Grok-2 models across a series of academic benchmarks that included reasoning, reading comprehension, math, science, and coding. Both Grok-2 and Grok-2 mini demonstrate significant improvements over our previous Grok-1.5 model. They achieve performance levels competitive to other frontier models in areas such as graduate-level science knowledge (GPQA), general knowledge (MMLU, MMLU-Pro), and math competition problems (MATH). Additionally, Grok-2 excels in vision-based tasks, delivering state-of-the-art performance in visual math reasoning (MathVista) and in document-based question answering (DocVQA).

Benchmark	Grok-1.5	Grok-2 mini^‡	Grok-2^‡	GPT-4 Turbo^*	Claude 3 Opus^†	Gemini Pro 1.5	Llama 3 405B	GPT-4o^*	Claude 3.5 Sonnet^†
GPQA	35.9%	51.0%	56.0%	48.0%	50.4%	46.2%	51.1%	53.6%	59.6%
MMLU	81.3%	86.2%	87.5%	86.5%	85.7%	85.9%	88.6%	88.7%	88.3%
MMLU-Pro	51.0%	72.0%	75.5%	63.7%	68.5%	69.0%	73.3%	72.6%	76.1%
MATH^§	50.6%	73.0%	76.1%	72.6%	60.1%	67.7%	73.8%	76.6%	71.1%
HumanEval^¶	74.1%	85.7%	88.4%	87.1%	84.9%	71.9%	89.0%	90.2%	92.0%
MMMU	53.6%	63.2%	66.1%	63.1%	59.4%	62.2%	64.5%	69.1%	68.3%
MathVista	52.8%	68.1%	69.0%	58.1%	50.5%	63.9%	—	63.8%	67.7%
DocVQA	85.6%	93.2%	93.6%	87.2%	89.3%	93.1%	92.2%	92.8%	95.2%

^* GPT-4-Turbo and GPT-4o scores are from the May 2024 release.
^† Claude 3 Opus and Claude 3.5 Sonnet scores are from the June 2024 release.
^‡ Grok-2 MMLU, MMLU-Pro, MMMU and MathVista were evaluated using 0-shot CoT.
^§ For MATH, we present maj@1 results.
^¶ For HumanEval, we report pass@1 benchmark scores.

Experience Grok with real-time information on 𝕏

Over the past few months, we've been continuously improving Grok on the 𝕏 platform. Today, we're introducing the next evolution of the Grok experience, featuring a redesigned interface and new features.

X App Grok page with Grok mini

Photo of a meme and Grok's explanation

Grok's response to a coding question

𝕏 Premium and Premium+ users will have access to two new models: Grok-2 and Grok-2 mini. Grok-2 is our state-of-the-art AI assistant with advanced capabilities in both text and vision understanding, integrating real-time information from the 𝕏 platform, accessible through the Grok tab in the 𝕏 app. Grok-2 mini is our small but capable model that offers a balance between speed and answer quality. Compared to its predecessor, Grok-2 is more intuitive, steerable, and versatile across a wide range of tasks, whether you're seeking answers, collaborating on writing, or solving coding tasks. In collaboration with Black Forest Labs, we are experimenting with their FLUX.1 model to expand Grok’s capabilities on 𝕏. If you are a Premium or Premium+ subscriber, make sure to update to the latest version of the 𝕏 app in order to beta test Grok-2.

Build with Grok using the Enterprise API

We are also releasing Grok-2 and Grok-2 mini to developers through our new enterprise API platform later this month. Our upcoming API is built on a new bespoke tech stack that allows multi-region inference deployments for low-latency access across the world. We offer enhanced security features such as mandatory multi-factor authentication (e.g. using a Yubikey, Apple TouchID, or TOTP), rich traffic statistics, and advanced billing analytics (incl. detailed data exports). We further offer a management API that allows you to integrate team, user, and billing management into your existing in-house tools and services. Join our newsletter to get notified when we launch later this month.

What is next?

Grok-2 and Grok-2 mini are being rolled out on 𝕏. We are very excited about their applications to a range of AI-driven features, such as enhanced search capabilities, gaining deeper insights on 𝕏 posts, and improved reply functions, all powered by Grok. Soon, we will release a preview of multimodal understanding as a core part of the Grok experience on 𝕏 and API.

Since announcing Grok-1 in November 2023, xAI has been moving at an extraordinary pace, driven by a small team with the highest talent density. We have introduced Grok-2, positioning us at the forefront of AI development. Our focus is on advancing core reasoning capabilities with our new compute cluster. We will have many more developments to share in the coming months. We are looking for individuals to join our small, focused team dedicated to building the most impactful innovations for the future of humanity. Apply to our positions here.

Series B funding round

Sun, 26 May 2024 00:00:00 +0000

xAI is pleased to announce...

Our Series B funding round of $6 billion with participation from key investors including Valor Equity Partners, Vy Capital, Andreessen Horowitz, Sequoia Capital, Fidelity Management & Research Company, Prince Alwaleed Bin Talal and Kingdom Holding, amongst others.

xAI has made significant strides over the past year. From the announcement of the company in July 2023, to the release of Grok-1 on X in November, to the recent announcements of the improved Grok-1.5 model with long context capability, to Grok-1.5V with image understanding, xAI’s model capabilities have improved rapidly. With the open-source release of Grok-1, xAI has opened doors for advancements in various applications, optimizations, and extensions of the model.

xAI will continue on this steep trajectory of progress over the coming months, with multiple exciting technology updates and products soon to be announced. The funds from the round will be used to take xAI’s first products to market, build advanced infrastructure, and accelerate the research and development of future technologies.

xAI is primarily focused on the development of advanced AI systems that are truthful, competent, and maximally beneficial for all of humanity. The company’s mission is to understand the true nature of the universe.

xAI is hiring for numerous roles and seeks talented individuals ready to join a small team focused on making a meaningful impact on the future of humanity. Those interested can apply today at x.ai/careers.

Grok-1.5 Vision Preview

Fri, 12 Apr 2024 00:00:00 +0000

Connecting the digital and physical worlds with our first multimodal model.

Announcing Grok-1.5

Thu, 28 Mar 2024 00:00:00 +0000

Introducing Grok-1.5, our latest model capable of long context understanding and advanced reasoning. Grok-1.5 will be available to our early testers and existing Grok users on the 𝕏 platform in the coming days.

By releasing the model weights and network architecture of Grok-1 two weeks ago, we presented a glimpse into the progress xAI had made up until last November. Since then, we have improved reasoning and problem-solving capabilities in our latest model, Grok-1.5.

Capabilities and Reasoning

One of the most notable improvements in Grok-1.5 is its performance in coding and math-related tasks. In our tests, Grok-1.5 achieved a 50.6% score on the MATH benchmark and a 90% score on the GSM8K benchmark, two math benchmarks covering a wide range of grade school to high school competition problems. Additionally, it scored 74.1% on the HumanEval benchmark, which evaluates code generation and problem-solving abilities.

Benchmark	Grok-1	Grok-1.5	Mistral Large	Claude 2	Claude 3 Sonnet	Gemini Pro 1.5	GPT-4	Claude 3 Opus
MMLU	73% 5-shot	81.3% 5-shot	81.2% 5-shot	75% 5-shot	79% 5-shot	83.7% 5-shot	86.4% 5-shot	86.8 5-shot
MATH	23.9% 4-shot	50.6% 4-shot	—	—	40.5% 4-shot	58.5% 4-shot	52.9% 4-shot	61% 4-shot
GSM8K	62.9 8-shot	90% 8-shot	81% 5-shot	88% 0-shot CoT	92.3% 0-shot CoT	91.7% 11-shot	92% 5-shot	95% 0-shot CoT
HumanEval	63.2% 0-shot	74.1% 0-shot	45.1% 0-shot	70% 0-shot	73% 0-shot	71.9% 0-shot	67% 0-shot	84.9% 0-shot

Long Context Understanding

A new feature in Grok-1.5 is the capability to process long contexts of up to 128K tokens within its context window. This allows Grok to have an increased memory capacity of up to 16 times the previous context length, enabling it to utilize information from substantially longer documents.

Furthermore, the model can handle longer and more complex prompts, while still maintaining its instruction-following capability as its context window expands. In the Needle In A Haystack (NIAH) evaluation, Grok-1.5 demonstrated powerful retrieval capabilities for embedded text within contexts of up to 128K tokens in length, achieving perfect retrieval results.

Grok-1.5 Infra

Cutting-edge Large Language Model (LLMs) research that runs on massive GPU clusters demands robust and flexible infrastructure. Grok-1.5 is built on a custom distributed training framework based on JAX, Rust, and Kubernetes. This training stack enables our team to prototype ideas and train new architectures at scale with minimal effort. A major challenge of training LLMs on large compute clusters is maximizing reliability and uptime of the training job. Our custom training orchestrator ensures that problematic nodes are automatically detected and ejected from the training job. We also optimized checkpointing, data loading, and training job restarts to minimize downtime in the event of a failure. If working on our training stack sounds interesting to you, apply to join the team.

Looking Ahead

Grok-1.5 will soon be available to early testers, and we look forward to receiving your feedback to help us improve Grok. As we gradually roll out Grok-1.5 to a wider audience, we are excited to introduce several new features over the coming days.

Note that the GPT-4 scores are taken from the March 2023 release. For MATH and GSM8K, we present maj@1 results. For HumanEval, we report pass@1 benchmark scores.

Open Release of Grok-1

Sun, 17 Mar 2024 00:00:00 +0000

We are releasing the base model weights and network architecture of Grok-1, our large language model. Grok-1 is a 314 billion parameter Mixture-of-Experts model trained from scratch by xAI.

This is the raw base model checkpoint from the Grok-1 pre-training phase, which concluded in October 2023. This means that the model is not fine-tuned for any specific application, such as dialogue.

We are releasing the weights and the architecture under the Apache 2.0 license.

To get started with using the model, follow the instructions at github.com/xai-org/grok.

Model Details

Base model trained on a large amount of text data, not fine-tuned for any particular task.
314B parameter Mixture-of-Experts model with 25% of the weights active on a given token.
Trained from scratch by xAI using a custom training stack on top of JAX and Rust in October 2023.

Announcing PromptIDE

Mon, 06 Nov 2023 00:00:00 +0000

The xAI PromptIDE is an integrated development environment for prompt engineering and interpretability research. It accelerates prompt engineering through an SDK that allows implementing complex prompting techniques and rich analytics that visualize the network's outputs. We use it heavily in our continuous development of Grok™.

We developed the PromptIDE to give transparent access to Grok-1, the model that powers Grok™, to engineers and researchers in the community. The IDE is designed to empower users and help them explore the capabilities of our large language models (LLMs) at pace. At the heart of the IDE is a Python code editor that - combined with a new SDK - allows implementing complex prompting techniques. While executing prompts in the IDE, users see helpful analytics such as the precise tokenization, sampling probabilities, alternative tokens, and aggregated attention masks.

The IDE also offers quality of life features. It automatically saves all prompts and has built-in versioning. The analytics generated by running a prompt can be stored permanently allowing users to compare the outputs of different prompting techniques. Finally, users can upload small files such as CSV files and read them using a single Python function from the SDK. When combined with the SDK's concurrency features, even somewhat large files can be processed quickly.

We also hope to build a community around the PromptIDE. Any prompt can be shared publicly at the click of a button. Users can decide if they only want to share a single version of the prompt or the entire tree. It's also possible to include any stored analytics when sharing a prompt.

The PromptIDE is available to members of our early access program. Below, you find a walkthrough of the main features of the IDE.

Thank you,
the xAI Team

Code editor & SDK

At the heart of the PromptIDE is a code editor and a Python SDK. The SDK provides a new programming paradigm that allows implementing complex prompting techniques elegantly. All Python functions are executed in an implicit context, which is a sequence of tokens. You can manually add tokens to the context using the prompt() function or you can use our models to generate tokens based on the context using the sample() function.

Code is executed locally using an in-browser Python interpreter that runs in a separate web worker. Multiple web workers can run at the same time, which means you can execute many prompts in parallel.

Complex prompting techniques can be implemented using multiple contexts within the same program. If a function is annotated with the @prompt_fn decorator, it is executed in its own, fresh context. The function can perform some operations independently of its parent context and pass the results back to the caller using the return statement. This programming paradigm enables recursive and iterative prompts with arbitrarily nested sub-contexts.

Concurrency

The SDK uses Python coroutines that enable processing multiple @prompt_fn-annotated Python functions concurrently. This can significantly speed up the time to completion - especially when working with CSV files.

User inputs

Prompts can be made interactive through the user_input() function, which blocks execution until the user has entered a string into a textbox in the UI. The user_input() function returns the string entered by the user, which cen then, for example, be added to the context via the prompt() function. Using these APIs, a chatbot can be implemented in just four lines of code:

python

await prompt(PREAMBLE)
while text := await user_input("Write a message"):
    await prompt(f"<|separator|>\n\nHuman: {text}<|separator|>\n\nAssistant:")
    await sample(max_len=1024, stop_tokens=["<|separator|>"], return_attention=True)

Files

Developers can upload small files to the PromptIDE (up to 5 MiB per file. At most 50 MiB total) and use their uploaded files in the prompt. The read_file() function returns any uploaded file as a byte array. When combined with the concurrency feature mentioned above, this can be used to implement batch processing prompts to evaluate a prompting technique on a variety of problems. The screenshot below shows a prompt that calculates the MMLU evaluation score.

Analytics

While executing a prompt, users see detailed per-token analytics to help them better understand the model's output. The completion window shows the precise tokenization of the context alongside the numeric identifiers of each token. When clicking on a token, users also see the top-K tokens after applying top-P thresholding and the aggregated attention mask at the token.

When using the user_input() function, a textbox shows up in the window while the prompt is running that users can enter their response into. The below screenshot shows the result of executing the chatbot code snippet listed above.

Finally, the context can also be rendered in markdown to improve legibility when the token visualization features are not required.

Announcing Grok

Fri, 03 Nov 2023 00:00:00 +0000

Grok is an AI modeled after the Hitchhiker’s Guide to the Galaxy, so intended to answer almost anything and, far harder, even suggest what questions to ask!

Grok is designed to answer questions with a bit of wit and has a rebellious streak, so please don’t use it if you hate humor!

A unique and fundamental advantage of Grok is that it has real-time knowledge of the world via the 𝕏 platform. It will also answer spicy questions that are rejected by most other AI systems.

Grok is still a very early beta product – the best we could do with 2 months of training – so expect it to improve rapidly with each passing week with your help.

Thank you,
the xAI Team

Why we are building Grok

At xAI, we want to create AI tools that assist humanity in its quest for understanding and knowledge.

By creating and improving Grok, we aim to:

Gather feedback and ensure we are building AI tools that maximally benefit all of humanity. We believe that it is important to design AI tools that are useful to people of all backgrounds and political views. We also want empower our users with our AI tools, subject to the law. Our goal with Grok is to explore and demonstrate this approach in public.
Empower research and innovation: We want Grok to serve as a powerful research assistant for anyone, helping them to quickly access relevant information, process data, and come up with new ideas.

Our ultimate goal is for our AI tools to assist in the pursuit of understanding.

The journey to Grok-1

The engine powering Grok is Grok-1, our frontier LLM, which we developed over the last four months. Grok-1 has gone through many iterations over this span of time.

After announcing xAI, we trained a prototype LLM (Grok-0) with 33 billion parameters. This early model approaches LLaMA 2 (70B) capabilities on standard LM benchmarks but uses only half of its training resources. In the last two months, we have made significant improvements in reasoning and coding capabilities leading up to Grok-1, a state-of-the-art language model that is significantly more powerful, achieving 63.2% on the HumanEval coding task and 73% on MMLU.

To understand the capability improvements we made with Grok-1, we have conducted a series of evaluations using a few standard machine learning benchmarks designed to measure math and reasoning abilities.

GSM8k: Middle school math word problems, (Cobbe et al. 2021), using the chain-of-thought prompt.

MMLU: Multidisciplinary multiple choice questions, (Hendrycks et al. 2021), provided 5-shot in-context examples.

HumanEval: Python code completion task, (Chen et al. 2021), zero-shot evaluated for pass@1.

MATH: Middle school and high school mathematics problems written in LaTeX, (Hendrycks et al. 2021), prompted with a fixed 4-shot prompt.

Benchmark	Grok-0 (33B)	LLaMa 2 70B	Inflection-1	GPT-3.5	Grok-1	Palm 2	Claude 2	GPT-4
GSM8k	56.8% 8-shot	56.8% 8-shot	62.9% 8-shot	57.1% 8-shot	62.9% 8-shot	80.7% 8-shot	88.0% 8-shot	92.0% 8-shot
MMLU	65.7% 5-shot	68.9% 5-shot	72.7% 5-shot	70.0% 5-shot	73.0% 5-shot	78.0% 5-shot	75.0% 5-shot + CoT	86.4% 5-shot
HumanEval	39.7% 0-shot	29.9% 0-shot	35.4% 0-shot	48.1% 0-shot	63.2% 0-shot	-	70% 0-shot	67% 0-shot
MATH	15.7% 4-shot	13.5% 4-shot	16.0% 4-shot	23.5% 4-shot	23.9% 4-shot	34.6% 4-shot	-	42.5% 4-shot

On these benchmarks, Grok-1 displayed strong results, surpassing all other models in its compute class, including ChatGPT-3.5 and Inflection-1. It is only surpassed by models that were trained with a significantly larger amount of training data and compute resources like GPT-4. This showcases the rapid progress we are making at xAI in training LLMs with exceptional efficiency.

Since these benchmarks can be found on the web and we can’t rule out that our models were inadvertently trained on them, we hand-graded our model (and also Claude-2 and GPT-4) on the 2023 Hungarian national high school finals in mathematics, which was published at the end of May, after we collected our dataset. Grok passed the exam with a C (59%), while Claude-2 achieved the same grade (55%), and GPT-4 got a B with 68%. All models were evaluated at temperature 0.1 and the same prompt. It must be noted that we made no effort to tune for this evaluation. This experiment served as a “real-life” test on a dataset our model was never explicitly tuned for.

Human-graded evaluation	Grok-0	GPT-3.5	Claude 2	Grok-1	GPT-4
Hungarian National High School Math Exam (May 2023)	37% 1-shot	41% 1-shot	55% 1-shot	59% 1-shot	68% 1-shot

We provide a summary of the important technical details of Grok-1 in the model card.

Engineering at xAI

At the frontier of deep learning research, reliable infrastructure must be built with the same care as datasets and learning algorithms. To create Grok, we built a custom training and inference stack based on Kubernetes, Rust, and JAX.

LLM training runs like a freight train thundering ahead; if one car derails, the entire train is dragged off the tracks, making it difficult to set upright again. There are a myriad of ways GPUs fail: manufacturing defects, loose connections, incorrect configuration, degraded memory chips, the occasional random bit flip, and more. When training, we synchronize computations across tens of thousands of GPUs for months on end, and all these failure modes become frequent due to scale. To overcome these challenges, we employ a set of custom distributed systems that ensure that every type of failure is immediately identified and automatically handled. At xAI, we have made maximizing useful compute per watt the key focus of our efforts. Over the past few months, our infrastructure has enabled us to minimize downtime and maintain a high Model Flop Utilization (MFU) even in the presence of unreliable hardware.

Rust has proven to be an ideal choice for building scalable, reliable, and maintainable infrastructure. It offers high performance, a rich ecosystem, and prevents the majority of bugs one would typically find in a distributed system. Given our small team size, infrastructure reliability is crucial, otherwise, maintenance starves innovation. Rust provides us with confidence that any code modification or refactor is likely to produce working programs that will run for months with minimal supervision.

We are now preparing for our next jump in model capabilities, which will require reliably coordinating training runs on tens of thousands of accelerators, running internet-scale data pipelines, and building new kinds of capabilities and tools into Grok. If that sounds exciting to you, apply to join the team here.

Research at xAI

We give Grok access to search tools and real-time information, but as with all the LLMs trained on next-token prediction, our model can still generate false or contradictory information. We believe that achieving reliable reasoning is the most important research direction to address the limitations of current systems. Here, we would like to highlight a few promising research directions we are most excited about at xAI:

Scalable oversight with tool assistance. Human feedback is essential. However, providing consistent and accurate feedback can be challenging, especially when dealing with lengthy code or complex reasoning steps. AI can assist with scalable oversight by looking up references from different sources, verifying intermediate steps with external tools, and seeking human feedback when necessary. We aim to make the most effective use of our AI tutors' time with the help of our models.
Integrating with formal verification for safety, reliability, and grounding. To create AI systems that can reason deeply about the real world, we plan to develop reasoning skills in less ambiguous and more verifiable situations. This allows us to evaluate our systems without human feedback or interaction with the real world. One major immediate goal of this approach is to give formal guarantees for code correctness, especially regarding formally verifiable aspects of AI safety.
Long-context understanding and retrieval. Training models for efficiently discovering useful knowledge in a particular context are at the heart of producing truly intelligent systems. We are working on methods that can discover and retrieve information whenever it is needed.
Adversarial robustness. Adversarial examples demonstrate that optimizers can easily exploit vulnerabilities in AI systems, both during training and serving time, causing them to make egregious mistakes. These vulnerabilities are long-standing weaknesses of deep learning models. We are particularly interested in improving the robustness of LLMs, reward models, and monitoring systems.
Multimodal capabilities. Currently, Grok doesn’t have other senses, such as vision and audio. To better assist users, we will equip Grok with these different senses that can enable broader applications, including real-time interactions and assistance.

We believe that AI holds immense potential for contributing significant scientific and economic value to society, so we will work towards developing reliable safeguards against catastrophic forms of malicious use. We believe in doing our utmost to ensure that AI remains a force for good.

If you share our optimism and want to contribute to our mission, apply to join the team here.

Early Access to Grok

We are offering a limited number of users in the United States to try out our Grok prototype and provide valuable feedback that will help us improve its capabilities before a wider release. You can join the Grok waitlist here. This release just represents the first step for xAI. Looking ahead, we have an exciting roadmap and will be rolling out new capabilities and features in the coming months.

Grok 4 Fast

Fri, 19 Sep 2025 00:00:00 +0000

We're thrilled to present Grok 4 Fast, our latest advancement in cost-efficient reasoning models. Built on xAI’s learnings from Grok 4, Grok 4 Fast delivers frontier-level performance across Enterprise and Consumer domains—with exceptional token efficiency. This model pushes the boundaries for smaller and faster AI, making high-quality reasoning accessible to more users and developers. Grok 4 Fast features state-of-the-art (SOTA) cost-efficiency, cutting-edge web and X search capabilities, a 2M token context window, and a unified architecture that blends reasoning and non-reasoning modes in one model.

Advancing Cost-Efficient Intelligence

Grok 4 Fast sets a new frontier in cost-efficient intelligence, outperforming Grok 3 Mini across reasoning benchmarks while slashing token costs.

Benchmark pass@1	Grok 4 Fast	Grok 4	Grok 3 Mini (High)	GPT-5 (High)	GPT-5 Mini (High)
GPQA Diamond	85.7%	87.5%	79.0%	85.7%	82.3%
AIME 2025 (no tools)	92.0%	91.7%	83.0%	94.6%	91.1%
HMMT 2025 (no tools)	93.3%	90.0%	74.0%	93.3%	87.8%
HLE (no tools)	20.0%	25.4%	11.0%	24.8%	16.7%
LiveCodeBench (Jan-May)	80.0%	79.0%	70.0%	86.8%	77.4%

We used large-scale reinforcement learning to maximize the intelligence density of Grok 4 Fast. In our evaluations, Grok 4 Fast achieves comparable performance to Grok 4 on benchmarks while using 40% fewer thinking tokens on average.

Intelligence Density

Maximum performance at minimum cost

AIME 2024 (no tools)

Score(%)

100%

Thinking tokens

28000

AIME 2025 (no tools)

Score(%)

100%

Thinking tokens

28000

HMMT 2025 (no tools)

Score(%)

100%

Thinking tokens

28000

GPQA Diamond

Score(%)

100%

Thinking tokens

28000

This 40% increase in Grok 4 Fast's token efficiency, combined with a significantly lower price per token, results in a 98% reduction in price to achieve the same performance on frontier benchmarks as Grok 4. As verified by an independent review from Artificial Analysis, Grok 4 Fast exhibits a state-of-the-art (SOTA) price-to-intelligence ratio compared to other publicly available models on the Artificial Analysis Intelligence Index.

Intelligence vs. Price

Artificial Analysis Intelligence Index

Cost to Run Intelligence Index(USD, Log Scale)

*All Claude models were benchmarked with Extended Thinking.

Native Tool Use with SOTA Search

Grok 4 Fast was trained end-to-end with tool-use reinforcement learning (RL). It excels at deciding when to invoke tools like code execution or web browsing.

For instance, Grok 4 Fast exhibits frontier agentic search capabilities, seamlessly browsing the web and X to augment queries with real-time data. It hops through links, ingests media (including images and videos on X), and synthesizes findings at light speed.

Benchmark pass@1	Grok 4 Fast	Grok 4	Grok 3 (No Reasoning)
BrowseComp	44.9%	43.0%	—
SimpleQA	95.0%	94.0%	82.0%
Reka Research Eval	66.0%	58.0%	37.0%
BrowseComp (zh)	51.2%	45.0%	10.8%
X Bench Deepsearch (zh)	74.0%	66.0%	27.0%
X Browse*	58.0%	53.2%	20.8%

*X Browse is an internal benchmark evaluating agent's multihop search and browsing capabilities on X.

Frontier of General Post-training

Grok 4 Fast also establishes a new cost-effective frontier on general domain. We are excited to share Grok 4 Fast’s result on LMArena, where it has been privately battle-testing on the Search and Text Arenas.

In LMArena's Search Arena, grok-4-fast-search (code name: menlo) claims #1 with 1163 Elo — a commanding margin of 17 over o3-search. Its superior reasoning efficiency and intelligence density enable it to surpass much larger models on real-world, search-related tasks.

In LMArena's Text Arena, grok-4-fast (code name: tahoe) ranks #8, performing on par with grok-4-0709 and highlighting its remarkable intelligence density. Notably, it significantly outperforms peers in its weight class, where all comparable size models rank 18th or below.

Here are some examples of Grok 4 Fast in action:

What is the maximum number of experience points possible in Path of Exile 2?

Thought for 24 seconds

Investigating game details

The request is about finding the maximum experience points in Path of Exile 2.

Path of Exile 2 is an action RPG, possibly still in early access as of September 10, 2025.

Searching for maximum experience points Path of Exile 2

reddit.com

polygon.com

turtlebeach.com

+2 others

The max level in Path of Exile 2 is 100, based on search results.

The question likely means the total XP needed to reach level 100.

Browsing https://www.poewiki.net/wiki/Experience

A recent search found the total XP to level 100 in Path of Exile is around 23 billion, per PoE Wiki.

Searching for total experience points to reach level 100 Path of Exile

reddit.com

pathofexile.fandom.com

pathofexile.com

+1 others

Refining search approach

Recent results suggest the total XP to level 100 might be over 4 billion, based on levels 91 and 97 data.

Browsing https://www.poewiki.net/wiki/Experience

Let's search for confirmation if PoE2 uses the same XP table as PoE1.

Searching for Path of Exile 2 experience requirements same as PoE1

reddit.com

pathofexile.com

pathofexile.fandom.com

+1 others

Earlier searches show max level 100 is the same for PoE1 and PoE2, with no mention of XP changes.

Searching for Path of Exile 2 total XP to level 100

reddit.com

pathofexile.com

pathofexile.fandom.com

+1 others

Since PoE2 shares the same game engine, it's likely the XP system is identical to PoE1.

Searching for Path of Exile 2 experience table level XP

pathofexile.fandom.com

reddit.com

poe2db.tw

+1 others

Found a site, PoE2DB, listing experience levels up to 100.

Browsing https://poe2db.tw/us/Experience

4,250,334,444

In Path of Exile 2, the maximum character level is 100, which requires a total of 4,250,334,444 experience points (XP) to reach from level 1. This value is the cumulative sum of XP needed for each level-up, based on the game's fixed experience table (identical to the original Path of Exile). After reaching level 100, additional XP can still be earned but does not grant further levels or meaningful progression.

To arrive at this total, the game's experience system uses a predefined table where each level requires progressively more XP. The "total XP" for level 100 is the sum of all XP thresholds from level 1 to 100. Here's how it's calculated step by step:

Understand the XP structure: Each level has an XP requirement to advance to the next (e.g., level 1 to 2 requires 525 XP). The total for a given level is the running sum of these requirements up to that point. Level 1 starts at 0 XP.
Reference the experience table: The official Path of Exile 2 data sources provide the full table. For example:
- Level 2: 525 total XP (525 to gain from level 1).
- Level 3: 1,760 total XP (1,235 to gain from level 2).
- ...and so on, up to level 100.
Sum the requirements: Add up the XP needed for all 99 level-ups (from 1 to 100). This can be computed programmatically or from the table's cumulative values, yielding 4,250,334,444.

Unified Model: Reasoning and Non-Reasoning

Previously, separate reasoning modes required distinct models. Grok 4 Fast introduces a unified architecture where reasoning (long chain-of-thought) and non-reasoning (quick responses) are handled by the same model weights, steered via system prompts. This unification reduces end-to-end latency as well as token costs, making Grok 4 Fast ideal for real-time applications.

In grok.com, this results in smooth transitions: responding instantly for simple queries or engaging in extended reasoning for complex ones. In the xAI API, developers can fine-tune this behavior, optimizing for speed or depth.

Grok 4 Fast in grok.com, iOS, and Android apps

Open

Grok.com

Open

Grok on iOS

Open

Grok on Android

Grok 4 Fast is available now for all users. In Fast and Auto modes, you will see a significant improvement in search and information seeking queries. Additionally, difficult queries in Auto mode will use Grok 4 Fast, which will provide a much faster experience without loss of quality. For the first time, all users, including free users, will have access to our latest model without restrictions, marking a step toward democratizing advanced AI.

Grok 4 Fast on OpenRouter, Vercel AI Gateway, and the xAI API

For a limited time, Grok 4 Fast will be available for free on OpenRouter and Vercel AI Gateway.

We're also rolling out Grok 4 Fast as two models: grok-4-fast-reasoning and grok-4-fast-non-reasoning, each with a 2M token context window. This allows developers to tune the amount of test-time compute applied to their use cases.

grok-4-fast-reasoning and grok-4-fast-non-reasoning are generally available via the xAI API according to the following pricing:

Token Type	<128k tokens	≥128k tokens
Input tokens	$0.20 / 1M	$0.40 / 1M
Output tokens	$0.50 / 1M	$1.00 / 1M
Cached input tokens	$0.05 / 1M

What's Next

We will continuously ship model improvements to Grok 4 Fast based on your feedback on x.com. Stay tuned for further integrations, including enhanced multimodal capabilities and agentic features.

Read the Grok 4 Fast model card here.

That's all for now - so long, and thanks for all the fish!

Expanding xAI for Government with GSA OneGov

Thu, 25 Sep 2025 00:00:00 +0000

Bringing accessible Frontier AI to the entire US Government

We’re excited to announce a significant expansion to our ‘xAI For Government’ offering – which we launched earlier this year in partnership with the General Services Administration (GSA). As a part of GSA’s OneGov Initiative, starting today, for the next 18 months, xAI For Government will be available to every federal government department, agency, or bureau, for a cost of just $0.42 for the period.

This new initiative brings our latest reasoning models, like Grok 4 and Grok 4 fast, into the hands of the government at an extremely affordable price. xAI is committed to bringing the best tools available to those working hard for our country, and this initiative lets government employees engage, experiment, and harness Frontier AI.

xAI is a strong supporter of President Trump’s AI Action Plan and this administration’s initiatives to bring the latest AI tools into the hands of federal government workers. We’re the only lab that has firmly committed to maintaining parity between the models that are available in the commercial space and tools available in regulated government environments.

"xAI has the most powerful AI compute and most capable AI models in the world. Thanks to President Trump and his administration, xAI’s frontier AI is now unlocked for every federal agency empowering the U.S. Government to innovate faster and accomplish its mission more effectively than ever before” said xAI cofounder and CEO Elon Musk. “We look forward to continuing to work with President Trump and his team to rapidly deploy AI throughout the government for the benefit of the country.”

xAI’s is committed to bring Frontier AI models to the government, but we won’t stop there. For AI to be maximally effective in the government, it needs to be implemented correctly, and the personnel using it need to be trained to use new AI-powered tools. To support this, xAI is committing a dedicated team of ‘Grok Engineers’ to assist and accelerate departments and agencies implementing AI tools. This is a unique offering and xAI is investing heavily not just in making the models accessible, but also in ensuring mission success.

“Widespread access to advanced AI models is essential to building the efficient, accountable government that taxpayers deserve — and to fulfilling President Trump’s promise that America will win the global AI race," said Federal Acquisition Service Commissioner Josh Gruenbaum. “We value xAI for partnering with GSA—and dedicating engineers—to accelerate the adoption of Grok to transform government operations.”

Learn more here or reach out to us directly.