Grok Chat
MCP server giving AI clients chat, vision, search and embeddings via the xAI Grok API
- Pricing
- Open source
- Platforms
- CLI, API, macOS
- Version
- 0.2.0
- Listing updated
- today
Similar MCP servers
About Grok Chat
Grok Chat is an open-source MCP server for ai models and llm tooling. It is available on CLI, API, macOS, Linux and Windows.
The README describes this as a Rust-built bridge that lets MCP-compatible clients call xAI's Grok models over stdio using JSON-RPC 2.0. Instead of wiring up the Grok HTTP API directly, a client gets named tools for chat completions, image analysis from a URL, grounded answers that pull in live web and X search, text embeddings, and a model listing.
The chat tool accepts multi-turn message history, system prompts, JSON-schema structured output and temperature settings. The README also notes that multi-agent model IDs are routed through the Responses API, with a reasoning_effort setting that changes how many agents run. A cached list_models call (five-minute cache) lets callers discover which model IDs are currently available.
It suits developers who already use Claude Desktop or another MCP client and want Grok available as a tool. Setup requires a Rust toolchain to build the binary and an xAI API key stored in a local config file.
Tools this server provides (5)
Taken from the project documentation; the running server may differ.
- chat
- Send a chat completion request to Grok with optional history, system prompt, structured output, model selection and multi-agent research.
- chat_with_vision
- Analyse an image given an image URL and text prompt.
- chat_with_search
- Chat with Grok using live web and/or X search to ground responses.
- embedding
- Generate text embeddings using a Grok embedding model.
- list_models
- List available Grok models and their IDs, cached for 5 minutes.
What it does
- Grounded web and X search
- Queries can be answered with live web search, X search, or both, chosen per request.
- Vision analysis from URLs
- Sends an image URL and a text prompt to a vision-capable Grok model, with a detail level setting.
- Structured JSON output
- Chat requests can include a JSON schema string to constrain the response format.
- Multi-turn conversations
- Accepts a message history array and an optional system prompt for context.
- Text embeddings
- Generates embeddings for a single string or an array of strings using a Grok embedding model.
- Model discovery
- Lists available Grok model IDs, cached for five minutes, so clients can pick a model.
Good for
- Give an MCP-enabled assistant access to Grok for general chat
- Ask questions answered with current web or X results
- Analyse images at a URL from inside an AI client
- Generate embeddings for search or retrieval pipelines through a tool call
Pricing
Open source
Free and MIT licensed; Grok API usage is billed by xAI using your own API key.
Grok Chat FAQ
What do I need to run it?
The README lists Rust (edition 2024) for building, and an xAI API key from console.x.ai placed in a config.toml file.
How does it communicate with clients?
It runs over stdio using JSON-RPC 2.0, as MCP servers do, and the README shows adding the built binary to Claude Desktop's MCP configuration.
What license is it under?
The README lists the MIT license.
Which model is used by default?
The chat, vision and search tools default to grok-4.3, and embedding defaults to grok-2-text-embedding.
Is Grok Chat free?
Yes. Grok Chat is open source, so you can use and self-host it for free.
What are the best alternatives to Grok Chat?
The closest alternatives to Grok Chat are Grok Image, Gpt Chat, MCP Aichat, Claude Chat and Crimson Crab MCP Template.
Where this comes from
- Description: written from the project's README and its MCP Registry entry.