Grok Community XDK

A community-maintained SDK (Standard Developer Kit) for XAI Grok models supporting usage with the Python Programming Language. This library provides convenient access to the xAI REST API from Python 3.11+ applications, with type definitions for all request parameters and response fields.

Installation

pip install xai-grok

Requirements

• Python 3.11 or higher
• pydantic
• requests

Usage

from xai_grok import Grok

client = Grok(
    api_key="your-api-key-here"
)

# Example: Create a chat completion
response = client.chat_completions(
    ChatRequest(
        messages=[
            {"role": "user", "content": "Tell me about AI"}
        ],
        model="model-name-here"
    )
)
print(response.choices[0].message.content)

Available Endpoints

API Key Operations

• api_key() - Retrieve information about the current API key

Chat and Completion Operations

• chat_completions(request: ChatRequest) - Create chat completions

  • Input: ChatRequest object containing messages and model settings
  • Throws: InvalidRequestError, IncompleteRequestError

• complete(request: CompleteRequest) - Generate completions

  • Input: CompleteRequest object with prompt and settings
  • Throws: InvalidRequestError, IncompleteRequestError

• completions(request: SampleRequest) - Alternative completion endpoint

  • Input: SampleRequest object
  • Throws: InvalidRequestError, IncompleteRequestError

Embedding Operations

• embedding_model(model_id: str) - Get details of a specific embedding model

  • Input: Model ID string
  • Throws: ModelNotFoundError

• embedding_models() - List all available embedding models

• embeddings(request: EmbeddingRequest) - Generate embeddings

  • Input: EmbeddingRequest object
  • Throws: InvalidRequestError, IncompleteRequestError

Language Model Operations

• language_model(model_id: str) - Get details of a specific language model

  • Input: Model ID string
  • Throws: ModelNotFoundError

• language_models() - List all available language models

Message Operations

• messages(request: MessageRequest) - Send messages
  • Input: MessageRequest object
  • Throws: InvalidRequestError, IncompleteRequestError

Model Management

• models() - List all available models

• model(model_id: str) - Get details of a specific model

  • Input: Model ID string
  • Throws: ModelNotFoundError

Text Operations

• tokenize_text(request: TokenizeTextRequest) - Tokenize input text
  • Input: TokenizeTextRequest object
  • Throws: InvalidRequestError

Request and Response Types

All request and response types are Pydantic models, providing type safety and validation. Refer to the schemas module for detailed type definitions.

Base URL

The API uses https://api.x.ai as the base URL for all endpoints.

Authentication

The API requires an authorized API key. Authentication-related errors are handled by specific error types:

• NoAPIKeyProvidedError: Raised when no API key is provided in the Authorization header
• InvalidAPIKeyProvidedError: Raised when an incorrect API key is provided

To avoid these errors:

1. Obtain a valid API key from console.x.ai
2. Include it in the client initialization:

client = Grok(api_key="your-api-key-here")

Note on Responses

All responses are parsed into their corresponding Pydantic models, providing type-safe access to response data. If the API response cannot be parsed into the expected type, a FailedToParseResponseError will be raised.