LlamaIndex + Aethon

Use Aethon as the LLM backend for your LlamaIndex RAG pipelines. Since Aethon is OpenAI-compatible, use LlamaIndex's built-in OpenAI class with a custom base URL.

Install

bash

pip install llama-index-llms-openai llama-index-embeddings-openai

Basic Usage

python

from llama_index.llms.openai import OpenAI

llm = OpenAI(
    api_base="https://api.aethon.ai/v1",
    api_key="your-aethon-key",
    model="llama3.1:8b",
)

response = llm.complete("Explain what a vector database is")
print(response.text)

Chat Interface

python

from llama_index.llms.openai import OpenAI
from llama_index.core.llms import ChatMessage

llm = OpenAI(
    api_base="https://api.aethon.ai/v1",
    api_key="your-aethon-key",
    model="llama3.1:8b",
)

messages = [
    ChatMessage(role="system", content="You are a helpful assistant."),
    ChatMessage(role="user", content="What is RAG?"),
]

response = llm.chat(messages)
print(response.message.content)

With Embeddings (RAG)

python

from llama_index.llms.openai import OpenAI
from llama_index.embeddings.openai import OpenAIEmbedding
from llama_index.core import Settings, VectorStoreIndex, SimpleDirectoryReader

# Configure Aethon as the LLM
Settings.llm = OpenAI(
    api_base="https://api.aethon.ai/v1",
    api_key="your-aethon-key",
    model="llama3.1:8b",
)

# Configure embeddings
Settings.embed_model = OpenAIEmbedding(
    api_base="https://api.aethon.ai/v1",
    api_key="your-aethon-key",
    model_name="nomic-embed-text",
)

# Build index and query
documents = SimpleDirectoryReader("./data").load_data()
index = VectorStoreIndex.from_documents(documents)
query_engine = index.as_query_engine()

response = query_engine.query("Summarize the key points")
print(response)

No wrapper package needed. LlamaIndex's built-in OpenAI integration works directly with Aethon. Just set api_base to Aethon's URL.

LangChain Guide →LiteLLM Guide →Open WebUI Guide →