Swarmauri Logo

Swarmauri LLM LeptonAI

Integration package for calling Lepton AI's hosted language and image generation models from Swarmauri agents. Ships LLM and image-gen adapters with synchronous, streaming, and asynchronous workflows that match Swarmauri conventions.

Features

Chat completion support for Lepton AI models (e.g., llama3-8b, mixtral-8x7b) with automatic usage tracking.
Streaming and async token generation for latency-sensitive experiences.
SDXL-based image generation with convenience helpers to save or display returned bytes.
Single configuration surface for model name, base URL, and API key; reuse the same credential for both text and image endpoints.

Prerequisites

Python 3.10 or newer.
A Lepton AI API key stored outside source control (environment variables or secret stores recommended).
Network access to *.lepton.run endpoints; the openai Python client is installed automatically as a dependency.

Installation

# pip
pip install swarmauri_llm_leptonai

# poetry
poetry add swarmauri_llm_leptonai

# uv (pyproject-based projects)
uv add swarmauri_llm_leptonai

Quickstart: Chat Completions

import os
from swarmauri_llm_leptonai import LeptonAIModel
from swarmauri_standard.conversations.Conversation import Conversation
from swarmauri_standard.messages.HumanMessage import HumanMessage

api_key = os.environ["LEPTON_API_KEY"]

conversation = Conversation()
conversation.add_message(HumanMessage(content="Summarize Swarmauri in two sentences."))

model = LeptonAIModel(api_key=api_key, name="llama3-8b")
response = model.predict(conversation=conversation)

print(response.get_last().content)
print("Tokens used", response.get_last().usage.total_tokens)

Async and Streaming

import asyncio
import os
from swarmauri_llm_leptonai import LeptonAIModel
from swarmauri_standard.conversations.Conversation import Conversation
from swarmauri_standard.messages.HumanMessage import HumanMessage

async def ask_async(prompt: str) -> None:
    convo = Conversation()
    convo.add_message(HumanMessage(content=prompt))

    model = LeptonAIModel(api_key=os.environ["LEPTON_API_KEY"], name="mixtral-8x7b")
    result = await model.apredict(conversation=convo)
    print(result.get_last().content)

def stream_story(prompt: str) -> None:
    convo = Conversation()
    convo.add_message(HumanMessage(content=prompt))

    model = LeptonAIModel(api_key=os.environ["LEPTON_API_KEY"])
    for token in model.stream(conversation=convo):
        print(token, end="", flush=True)

# asyncio.run(ask_async("Draft a product announcement."))
# stream_story("Write a haiku about distributed agents.")

Generate Images with SDXL

import os
from pathlib import Path
from swarmauri_llm_leptonai import LeptonAIImgGenModel

img_model = LeptonAIImgGenModel(api_key=os.environ["LEPTON_API_KEY"], model_name="sdxl")

prompt = "A cyberpunk skyline at blue hour in watercolor style"
image_bytes = img_model.generate_image(prompt=prompt, width=768, height=512)

output = Path("leptonai_cyberpunk.png")
img_model.save_image(image_bytes, output.as_posix())

# Display in a notebook or desktop environment
# img_model.display_image(image_bytes)

Operational Tips

Models are invoked via https://<model>.lepton.run/api/v1/; updating name on LeptonAIModel switches endpoints without altering the client setup.
Streaming responses emit usage data at stream completion; consume the generator fully before inspecting conversation.get_last().usage.
Respect Lepton AI rate limits—add retries with exponential backoff or queue requests during traffic spikes.
Store API keys securely and rotate them regularly; avoid hard-coding credentials in notebooks or scripts.
Large image generations may take longer and consume more credits; adjust width, height, steps, and guidance_scale to balance quality versus latency.

Want to help?

If you want to contribute to swarmauri-sdk, read up on our guidelines for contributing that will help you get started.

Metadata

Release files for swarmauri_llm_leptonai 0.9.3

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for swarmauri_llm_leptonai 0.9.3
File	Size	Uploaded
swarmauri_llm_leptonai-0.9.3.tar.gz	10.2 kB	Mar 24, 2026	Details

Built distribution (wheel)

Table of built distributions (wheels) for swarmauri_llm_leptonai 0.9.3
File	Interpreter	ABI	Platform	Reset
swarmauri_llm_leptonai-0.9.3-py3-none-any.whl 11.7 kB Mar 24, 2026	Python 3	none	any	Details

Total release size: 21.8 kB

Release files / swarmauri_llm_leptonai-0.9.3.tar.gz

Download URL	swarmauri_llm_leptonai-0.9.3.tar.gz
Size	10.2 kB
Tags	Source
SHA-256 checksum How to use checksums	`650c311b3629fba9be23533d5c6af0203293f104d9602eeefcd04e8b2857347f`
BLAKE2b-256 checksum How to use checksums	`45346cc0c2b5239ab278a0f4781e4c3d9faa1b9dbe8ba089ad31e2fa2e46772a`
Upload date	Mar 24, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`uv/0.11.0 {"installer":{"name":"uv","version":"0.11.0","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Ubuntu","version":"24.04","id":"noble","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":true}`

Release files / swarmauri_llm_leptonai-0.9.3-py3-none-any.whl

Download URL	swarmauri_llm_leptonai-0.9.3-py3-none-any.whl
Size	11.7 kB
Tags	Python 3
SHA-256 checksum How to use checksums	`de6f1149d04d1a0895a5c938f0ace8dbab69540e695834d1cfd29bdc4372d07b`
BLAKE2b-256 checksum How to use checksums	`cb018c4d07372e8e703e84625367af0849c0d9cf7af10d7b80f0274ce5a089fb`
Upload date	Mar 24, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`uv/0.11.0 {"installer":{"name":"uv","version":"0.11.0","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Ubuntu","version":"24.04","id":"noble","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":true}`