Chat AI xAI

Grok

Real-time AI assistant with X-integrated search, coding, and multimodal generation

Visit Grok →

What it does

Grok is a real-time AI assistant developed by xAI. It combines chat, web search, X search, reasoning, coding, voice, image understanding, image generation, and video generation into one product family. Available via grok.com, mobile apps, X integration, API, and the Grok Build CLI.

As of May 2026, xAI positions Grok 4.3 as its primary API model. It supports 1M-token context, text + image input, configurable reasoning, function calling, structured outputs, and Remote MCP Tools. For coding, Grok Build 0.1 powers a terminal-native agentic coding experience. For creative work, Grok Imagine API provides a separate image and video generation layer.

Grok’s biggest differentiator is its real-time information layer connected to X and its attempt to unify chat, search, voice, image, video, and coding under one API.

Models

ModelContextInputOutput
Grok 4.31M tokens$1.25/M$2.50/M
Grok Build 0.1256K tokens$1/M$2/M
Grok 4.1 Fast2M tokens$0.20/M$0.50/M

Grok 4.3 is xAI’s current flagship model, focused on strong agentic tool calling, low-hallucination performance, configurable reasoning, and vision. Cached input is priced at $0.20/M across all current models.

Pricing

  • Free ($0/mo) — real-time web + X search, voice mode, connectors
  • SuperGrok Lite ($10/mo) — cheapest paid plan with higher limits than Free
  • SuperGrok ($30/mo) — Grok 4, higher rate limits, Expert mode, image/video generation
  • X Premium+ ($40/mo) — Grok access bundled with X platform features
  • SuperGrok Heavy ($300/mo) — highest individual tier, Grok 4 Heavy, early feature access
  • Business / Enterprise — team management, consolidated billing, SSO, custom pricing

On the API side, Grok 4.3 is priced at $1.25/M input and $2.50/M output. Image and video generation use image- or second-based pricing rather than token pricing.

Capabilities

  • Real-time web and X search
  • Long-context reasoning up to 1M tokens
  • Vision: image understanding
  • Function calling and structured outputs
  • Remote MCP Tools for external system connections
  • Code execution, file/collection search, and server-side tools
  • Grok Build CLI with agentic coding, skills, plugins, hooks, and MCP servers
  • Imagine API for image and video generation/editing

Strengths

  • Strong real-time social/news context through X integration
  • Aggressive Grok 4.3 API pricing compared with many frontier models
  • Combines chat, search, voice, image, video, and coding in one ecosystem
  • Remote MCP Tools make Grok usable inside agent workflows
  • Grok Build offers a Claude Code / Cursor-like coding-agent experience
  • OpenAI-compatible API path makes migration easier

Weaknesses

  • Consumer subscription limits are less transparent than API pricing
  • Past controversies around safety and public response behavior
  • Smaller third-party marketplace than GPT Store, Claude Skills, or Cursor ecosystems
  • Many of the best features may require SuperGrok, X Premium+, or early access
  • Model and feature availability can vary by region, account, and console access
  • Creative and coding ecosystems are growing fast but still have fewer mature standards

Ecosystem

Grok’s ecosystem has three main parts: Grok Build, Remote MCP Tools, and Grok Imagine.

Grok Build is a terminal-native coding agent. It supports skills, plugins, hooks, AGENTS.md, MCP servers, subagents, plan mode, code review, sandboxed execution, and headless mode. This gives Grok a developer experience closer to Claude Code and Cursor.

Remote MCP Tools let Grok connect to external MCP servers. This allows Grok to work with third-party tools or custom internal MCP servers for external system integration.

Grok Imagine is xAI’s image and video generation/editing layer. Priced separately from chat models, it expands Grok from a chat assistant into a multimodal creation platform.