Plugin for LLM providing access to Grok models using the xAI API

Install this plugin in the same environment as LLM:

llm install llm-grok

First, obtain an API key from xAI.

Configure the key using the llm keys set grok command:

llm keys set grok
# Paste your xAI API key here

You can also set it via environment variable:

export XAI_API_KEY="your-api-key-here"

You can now access the Grok model. Run llm models to see it in the list.

To run a prompt through grok-4-fast (default model):

llm -m grok-4-fast 'What is the meaning of life, the universe, and everything?'

To start an interactive chat session:

llm chat -m grok-4-fast

Example chat session:

Chatting with grok-4-fast
Type 'exit' or 'quit' to exit
Type '!multi' to enter multiple lines, then '!end' to finish
> Tell me a joke about programming

To use a system prompt to give Grok specific instructions:

cat example.py | llm -m grok-4-fast -s 'explain this code in a humorous way'

To set your default model:

llm models default grok-3-mini-latest
# Now running `llm ...` will use `grok-3-mini-latest` by default

The following Grok models are available:

• grok-4-latest
• grok-4-fast (default)
• grok-4-fast-reasoning-latest
• grok-4-fast-non-reasoning-latest
• grok-code-fast-1
• grok-3-latest
• grok-3-mini-fast-latest
• grok-3-mini-latest
• grok-3-fast-latest
• grok-2-latest
• grok-2-vision-latest

You can check the available models using:

llm grok models

The grok models accept the following options, using -o name value syntax:

• -o temperature 0.7: The sampling temperature, between 0 and 1. Higher values like 0.8 increase randomness, while lower values like 0.2 make the output more focused and deterministic.
• -o max_completion_tokens 100: Maximum number of tokens to generate in the completion (includes both visible tokens and reasoning tokens).

All Grok models support live search functionality to access real-time information:

• -o search_mode auto: Live search mode. Options: auto, on, off (default: disabled)
• -o max_search_results 20: Maximum number of search results to consider (default: 20)
• -o return_citations true: Whether to return citations for search results (default: true)
• -o search_from_date 2025-01-01: Start date for search results in ISO8601 format (YYYY-MM-DD)
• -o search_to_date 2025-01-15: End date for search results in ISO8601 format (YYYY-MM-DD)

• -o excluded_x_handles "@spam_account,@another": Comma-separated list of X handles to exclude (max 10)
• -o included_x_handles "@elonmusk,@openai": Comma-separated list of X handles to include (cannot be used with excluded_x_handles)
• -o post_favorite_count 100: Minimum number of favorites for X posts to be included
• -o post_view_count 1000: Minimum number of views for X posts to be included

Basic usage with options:

llm -m grok-4-fast -o temperature 0.2 -o max_completion_tokens 50 'Write a haiku about AI'

Using live search to get current information:

llm -m grok-4-fast -o search_mode on 'What are the latest developments in AI today?'

Searching with date constraints:

llm -m grok-4-fast -o search_mode on -o search_from_date 2025-01-01 -o search_to_date 2025-01-15 'What happened in AI this month?'

Filtering X posts by engagement:

llm -m grok-4-fast -o search_mode on -o post_favorite_count 1000 -o post_view_count 10000 'Show me popular AI discussions on X'

Excluding specific X accounts:

llm -m grok-4-fast -o search_mode on -o excluded_x_handles "@spam_account" 'Latest AI news from X'

To set up this plugin locally, first checkout the code. Then create a new virtual environment:

git clone https://github.com/hiepler/llm-grok.git
cd llm-grok
python3 -m venv venv
source venv/bin/activate

Now install the dependencies and test dependencies:

pip install -e '.[test]'

To run the tests:

pytest

List available Grok models:

llm grok models

This plugin uses the xAI API. For more information about the API, see:

• xAI API Documentation

Contributions are welcome! Please feel free to submit a Pull Request.

Apache License 2.0