Swarmauri xAI LLM

swarmauri_llm_xai provides provider-native Swarmauri adapters for xAI's Grok chat-completions API.

Features

• XAIModel directly implements the Swarmauri LLM base contract.
• XAIToolModel directly implements the tool-LLM base contract.
• Bearer-token authentication with credentials excluded from serialized state.
• Sync, async, incremental streaming, batch, and function-calling workflows.
• Optional model discovery through xAI's language-model catalog.
• Explicit provider, HTTP client, and retry-policy composition.
• Conservative chat retries that do not replay ambiguous connection failures.

Installation

uv add swarmauri_llm_xai

pip install swarmauri_llm_xai

Usage

import os

from swarmauri_llm_xai import XAIModel
from swarmauri_standard.conversations.Conversation import Conversation
from swarmauri_standard.messages.HumanMessage import HumanMessage

conversation = Conversation(
    messages=[HumanMessage(content="Explain orbital mechanics briefly.")]
)
model = XAIModel(api_key=os.environ["XAI_API_KEY"], name="grok-4.3")
model.predict(conversation)
print(conversation.get_last().content)

For explicit composition:

from swarmauri_llm_xai import XAIChatRetryPolicy, XAIModel
from swarmauri_provider_xai import XAIProvider
from swarmauri_transport_httpx import HttpxClient

provider = XAIProvider(api_key=os.environ["XAI_API_KEY"])
model = XAIModel(
    provider=provider,
    http_client=HttpxClient(base_url=provider.base_url),
    retry_policy=XAIChatRetryPolicy(max_attempts=3),
)

Use XAIToolModel with a Swarmauri Toolkit for function calling. Set discover_models=True to populate allowed model IDs from the authenticated xAI language-model catalog.

See the xAI inference reference, chat endpoint, and function-calling guide.