Grok
AI model by xAI for conversation and discovery, accessible on web, iOS, and Android with a developer API.
Last verified:
What is Grok?
Grok is a family of large language models (LLMs) developed by xAI, designed as a maximally truth-seeking AI that provides insightful, unfiltered truths about the universe. Grok is accessible through grok.com, iOS and Android apps, and directly within X.com and X apps. The same models power consumer-facing services including Grok.com, mobile apps, and the Grok in X experience.
The xAI API enables developers to programmatically integrate Grok models into their applications, providing building blocks to create new AI experiences. Key features include native tool use (autonomous web search, code interpreter execution), real-time search integration across X, the web, and news sources, structured outputs for JSON-formatted responses, vision capabilities for image understanding, voice mode with natural conversation, image generation through Grok Imagine, and video generation. Grok 4 offers a 256,000 token context window, multi-agent reasoning, DeepSearch for comprehensive web research, and Big Brain Mode for extended reasoning on complex problems.
Grok is ideal for developers building AI applications, traders needing real-time data, research teams requiring advanced reasoning, government customers for national security and science applications, and everyday users wanting a curious, truthful AI assistant with real-time X data access. Grok For Government provides frontier AI products specifically for U.S. government customers including national security, federal, state, and local agencies.
The platform supports multimodal interaction across text, voice, image, and video, with multilingual audio processing capabilities. Grok Skills enable persistent custom expertise that models retain across conversations, while the Responses API supports parallel tool calls, up to 128 tools per request, and a 1 million token context window for multi-step agentic tasks.
Grok pricing
Pricing model: Free
Free tier: Access to Grok 4 and Grok 4.1 at basic usage levels (approximately 10 prompts per two-hour window), standard web and X search integration, basic image generation through Grok Imagine, and voice mode with limited session length. SuperGrok: $30/month or $300/year (about $25/month), includes unlimited usage, Grok 4 access, DeepSearch, enhanced reasoning modes, higher image generation limits, 256,000 tokens of memory per conversation, and early access to experimental features. SuperGrok Heavy: $300/month, provides full Grok 4 Heavy access with multi-agent reasoning, maximum rate limits everywhere, priority on every model, fastest access to high-compute models, extended multimodal features, reserved GPU slots during peak times, and full Grok Imagine video generation. API pricing: Available through xAI API Endpoint (https://api.x.ai) with credits required after account creation. Grok For Government: Available first to United States Government customers with custom pricing for national security, federal, state, and local agencies.
Grok pros
- Maximally truth-seeking AI with unfiltered responses
- Real-time data access from X (Twitter) platform
- Native tool use for autonomous web search and code execution
- 256,000 token context window in Grok 4
- Structured outputs for precise JSON-formatted responses
- Vision capabilities for image understanding and analysis
- Voice mode with natural speak-and-listen conversation
- Image generation through Grok Imagine feature
- Video generation including 6-second videos with synchronized sound
- DeepSearch for comprehensive multi-step web research
- Big Brain Mode for extended reasoning on complex problems
- Multi-agent reasoning in Grok 4 Heavy
- Grok Skills for persistent custom expertise across conversations
- OpenAI-compatible API for easy developer integration
- 1 million token context window for agentic tasks
- Parallel tool calls supporting up to 128 tools per request
- Live search API for real-time data across X, web, and news
- Multilingual audio processing capabilities
- Priority routing and maximum rate limits in SuperGrok Heavy
- Early access to experimental features before public release
- Available on web, iOS, Android, and within X platform
- Grok For Government for classified and restricted environments
- SOC 2 Type 2, GDPR, and CCPA compliance certifications
- Custom AI-powered applications for healthcare and science use cases
Grok cons
- Free tier limited to approximately 10 prompts per two-hour window
- SuperGrok Heavy tier costs $300/month, unusually expensive
- Free tier does not include DeepSearch or Big Brain Mode
- Image generation has strict daily caps on free tier
- Voice mode has limited session length on free tier
- Grok 4 Heavy access restricted to premium subscribers only
- Some advanced features require SuperGrok subscription at $30/month
- Free tier has standard response speed without priority queuing
- Extended memory and higher context windows not available on free tier
Frequently asked questions about Grok
What is Grok and what makes it different from other AI assistants?
Grok is a family of large language models developed by xAI, designed as a maximally truth-seeking AI that provides insightful, unfiltered truths about the universe. Named after the Hitchhiker's Guide to the Galaxy, Grok stands out with real-time data access from X (Twitter), native tool use for autonomous web search and code execution, and a truth-seeking approach that avoids filtering responses.
How do I access Grok and where can I use it?
Grok is accessible on grok.com, through iOS and Android apps, and directly within X.com and X apps. The same models power all consumer-facing services. Developers can also access Grok programmatically through the xAI API at https://api.x.ai, which is fully compatible with the OpenAI SDK.
What models are available in the Grok family?
The Grok lineup includes Grok 4 (the most intelligent model in the world with native tool use and real-time search), Grok 4 Heavy (the most powerful version with multi-agent reasoning), Grok 3, and Grok 3 mini. Grok 4 features a 256,000 token context window, while the API supports up to 1 million token context windows for agentic tasks.
What is the difference between SuperGrok and SuperGrok Heavy?
SuperGrok at $30/month provides unlimited Grok 4 access, DeepSearch, enhanced reasoning modes, higher image generation limits, and 256,000 tokens of memory. SuperGrok Heavy at $300/month adds full Grok 4 Heavy access with multi-agent reasoning, maximum rate limits everywhere, priority on every model, fastest access to high-compute models, reserved GPU slots during peak times, and full video generation capabilities.
What features does the free tier include and what are its limitations?
The free tier includes access to Grok 4 and Grok 4.1 at approximately 10 prompts per two-hour window, standard web and X search integration, basic image generation through Grok Imagine, and voice mode with limited session length. Limitations include no DeepSearch, no Big Brain Mode, strict daily image generation caps, shorter voice session limits, standard response speed without priority, and no access to extended memory or higher context windows.
What is DeepSearch and how does it work?
DeepSearch is Grok's deep research mode that conducts multi-step web research, synthesizes findings from multiple sources, and produces detailed reports. It combines X data with broader web sources, scanning approximately 100 pages and X profiles to create comprehensive, up-to-date summaries. The feature was upgraded to DeeperSearch in March 2025, adding reasoning layers to reconcile conflicting data points and weigh source credibility.
What is Big Brain Mode and when should I use it?
Big Brain Mode is Grok's extended thinking capability for complex problems, similar to OpenAI's o1/o3 reasoning models. It allows Grok to work through multi-step logic, math, and coding problems with visible reasoning chains, allocating additional GPU resources dynamically for deeper exploration of candidate solutions. It's available on Grok 3 and newer models but restricted to SuperGrok subscribers.
What vision and multimedia capabilities does Grok have?
Grok features vision capabilities for image understanding and analysis (Grok Vision), including object identification, text reading from images, and scene description. It supports image generation through Grok Imagine (powered by xAI's Aurora model), video generation including 6-second videos with synchronized sound from text prompts or image-to-video conversion, and multilingual audio processing with real-time search in voice mode.
How does the xAI API work for developers?
The xAI API is a toolkit for developers to integrate Grok models into applications, providing building blocks for new AI experiences. Developers create an xAI account, generate an API key via the API Keys Page, load credits, and make requests using the xAI Python SDK or OpenAI-compatible format. The API supports tool calling with web_search, x_search, and code_interpreter types, structured outputs via response_format parameter, and is compatible with both xai-sdk and OpenAI SDK.
What is Grok For Government and what does it offer?
Grok For Government is a suite of frontier AI products available first to United States Government customers, including national security, federal, state, and local agencies. It includes Grok 4 with strong reasoning capabilities, Deep Search, Tool Use, custom AI-powered applications for national security, healthcare, and fundamental science, models in classified and restricted environments, security-cleared engineering support, and custom versions built for specific missions.