swarmauri_llm_ai21 0.11.0.dev1

Provider-specific Swarmauri import package for AI21 Studio Jamba chat completions with sync, async, streaming, batch, and usage workflows.

Swarmauri AI21 LLM

swarmauri_llm_ai21 provides the AI21StudioModel import path for Swarmauri applications that use AI21 Studio Jamba chat models. The package re-exports the maintained implementation from swarmauri_standard, giving projects a provider-specific dependency while preserving Swarmauri conversation, message, usage, streaming, and batch workflows.

Why Swarmauri AI21 LLM?

Use this package when you want to install only the AI21-facing LLM provider surface for a Swarmauri application. AI21StudioModel formats Swarmauri conversations for AI21 Studio chat completions, sends bearer-token authenticated HTTP requests, supports sync and async calls, streams token deltas, records usage data when available, and keeps model selection inside the Swarmauri LLMBase interface.

FAQ

Q: Which AI21 API does this adapter call?

A: The adapter posts to https://api.ai21.com/studio/v1/chat/completions, AI21 Studio's chat completions endpoint used by Jamba chat models.

Q: Which model names are configured?

A: The current allowed model list is jamba-large, jamba-mini, jamba-large-1.7, and jamba-mini-1.7. The default model name is jamba-large.

Q: Does this package contain a separate AI21 implementation?

A: No. It re-exports AI21StudioModel from swarmauri_standard.llms.AI21StudioModel so applications can depend on the provider package while using the shared implementation.

Q: What credential is required?

A: Pass an AI21 Studio API key with AI21StudioModel(api_key="..."). The repository tests look for AI21STUDIO_API_KEY when running live provider checks.

Features

• Provider-specific import: from swarmauri_llm_ai21 import AI21StudioModel.
• Swarmauri Conversation, HumanMessage, SystemMessage, and AgentMessage integration.
• Sync predict() and async apredict() chat completion calls.
• Sync stream() and async astream() token streaming.
• Sync batch() and async abatch() helpers for multiple conversations.
• Usage metadata mapping to Swarmauri UsageData when provider usage data is returned.
• Retry handling for rate-limit and overloaded-provider status codes.
• Python 3.10, 3.11, 3.12, 3.13, and 3.14 support.

Prerequisites

• An AI21 Studio account and API key.
• Network access to api.ai21.com.
• Swarmauri conversations built from Swarmauri message classes.

Installation

Install with uv:

uv add swarmauri_llm_ai21

Install with pip:

pip install swarmauri_llm_ai21

Usage

Run a single chat completion:

import os

from swarmauri_llm_ai21 import AI21StudioModel
from swarmauri_standard.conversations.Conversation import Conversation
from swarmauri_standard.messages.HumanMessage import HumanMessage
from swarmauri_standard.messages.SystemMessage import SystemMessage

conversation = Conversation()
conversation.add_message(SystemMessage(content="Answer with concise technical prose."))
conversation.add_message(HumanMessage(content="What is a Swarmauri component?"))

model = AI21StudioModel(
    api_key=os.environ["AI21STUDIO_API_KEY"],
    name="jamba-mini",
)
result = model.predict(conversation=conversation, temperature=0.3, max_tokens=256)

print(result.get_last().content)

Stream a response:

import os

from swarmauri_llm_ai21 import AI21StudioModel
from swarmauri_standard.conversations.Conversation import Conversation
from swarmauri_standard.messages.HumanMessage import HumanMessage

conversation = Conversation()
conversation.add_message(HumanMessage(content="Write a short release note."))

model = AI21StudioModel(api_key=os.environ["AI21STUDIO_API_KEY"])

for token in model.stream(conversation=conversation, max_tokens=160):
    print(token, end="")

Use the async API:

import os
import asyncio

from swarmauri_llm_ai21 import AI21StudioModel
from swarmauri_standard.conversations.Conversation import Conversation
from swarmauri_standard.messages.HumanMessage import HumanMessage


async def main() -> None:
    conversation = Conversation()
    conversation.add_message(HumanMessage(content="Summarize AI21 Jamba in one line."))

    model = AI21StudioModel(api_key=os.environ["AI21STUDIO_API_KEY"])
    result = await model.apredict(conversation=conversation)

    print(result.get_last().content)


asyncio.run(main())

Related Packages

LLM provider packages:

• swarmauri_llm_anthropic
• swarmauri_llm_cerebras
• swarmauri_llm_cohere
• swarmauri_llm_gemini
• swarmauri_llm_groq
• swarmauri_llm_mistral
• swarmauri_llm_openai

Foundational packages:

• swarmauri_core defines core interfaces.
• swarmauri_base provides base component classes.
• swarmauri_standard contains the shared AI21StudioModel implementation.
• swarmauri provides namespace imports and plugin discovery.

Best Practices

• Read the API key from environment variables or a secret manager.
• Set name explicitly so deployments do not change behavior when defaults change.
• Use async methods for high-concurrency applications.
• Keep prompts and system messages in Swarmauri Conversation objects for provider portability.
• Capture UsageData for cost and latency monitoring where provider responses include token usage.

License

Apache-2.0