LlamaIndex

0.14.20 · active · verified Thu Apr 09

LlamaIndex is a data framework for LLM applications, providing tools to ingest, structure, and access private or domain-specific data with large language models. It facilitates building RAG (Retrieval Augmented Generation) applications, agents, and more. The current version is 0.14.20, with a rapid release cadence that often includes new integrations and improvements across its modular ecosystem.

Warnings

breaking LlamaIndex underwent a major architectural refactor in version 0.10.0. The `ServiceContext` class was deprecated and replaced by the global `Settings` object for configuring LLMs, embedding models, and other components. Many core classes moved or were renamed.
Fix: Migrate your code to use `llama_index.core.Settings` for configuration. For example, instead of `ServiceContext.from_defaults(llm=..., embed_model=...)`, use `Settings.llm = ...` and `Settings.embed_model = ...`. Refer to the official migration guide for 0.10.0.
breaking The library transitioned to a modular package structure in 0.10.0. While `llama-index` is a metapackage, specific LLM, embedding, vector store, and other integrations are now separate packages (e.g., `llama-index-llms-openai`, `llama-index-embeddings-openai`).
Fix: Explicitly install the required integration packages using `pip install` (e.g., `pip install llama-index-llms-openai`). Imports also changed to reflect this modularity (e.g., `from llama_index.llms.openai import OpenAI`).
deprecated Support for Python 3.9 was deprecated and eventually removed from various sub-packages.
Fix: Upgrade your Python environment to Python 3.10 or higher. The library officially requires `>=3.10` and `<4.0`.
gotcha LlamaIndex often requires API keys or credentials for external services (e.g., OpenAI, Anthropic, various vector databases). It does not manage these directly.
Fix: Ensure relevant API keys (e.g., `OPENAI_API_KEY`, `ANTHROPIC_API_KEY`) are set as environment variables or passed directly to the client constructors where applicable.
gotcha Due to rapid development and frequent releases, minor versions can sometimes introduce breaking changes or significant refactors without a new major version number. Always check release notes for your specific version.
Fix: Pin your `llama-index` dependencies to specific versions, or thoroughly test your application when updating. Review the GitHub release notes before upgrading.

Install

pip install llama-index Base installation
pip install llama-index-llms-openai llama-index-embeddings-openai Common integrations (e.g., OpenAI)

Imports

VectorStoreIndex

from llama_index.core import VectorStoreIndex

SimpleDirectoryReader

from llama_index.core import SimpleDirectoryReader

Settings
```
from llama_index.core import Settings
```
`ServiceContext` was deprecated in LlamaIndex 0.10.0 and replaced by the global `Settings` object for configuring LLM, embedding model, and other core components.
OpenAI
```
from llama_index.llms.openai import OpenAI
```
As of LlamaIndex 0.10.0, LLM and Embedding models are imported from their specific integration packages (e.g., `llama_index.llms.openai`).
OpenAIEmbedding
```
from llama_index.embeddings.openai import OpenAIEmbedding
```
As of LlamaIndex 0.10.0, LLM and Embedding models are imported from their specific integration packages (e.g., `llama_index.embeddings.openai`).

Quickstart

This quickstart demonstrates how to load a document, create a vector index, and query it using the default OpenAI LLM and embedding models. It highlights the use of the `Settings` object for configuration, which replaced `ServiceContext` in LlamaIndex 0.10.0+.

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

# Ensure you have OPENAI_API_KEY set in your environment variables
# For quick testing, a dummy key is used, but a real key is needed for actual API calls.
os.environ["OPENAI_API_KEY"] = os.environ.get("OPENAI_API_KEY", "sk-DUMMY")

# Create a dummy data directory and file for the example
if not os.path.exists("data"): 
    os.makedirs("data")
with open("data/sample_doc.txt", "w") as f:
    f.write("LlamaIndex is a data framework for building LLM applications.")
    f.write("It helps connect custom data sources to large language models.")

try:
    # Configure the global Settings object (replaces ServiceContext)
    Settings.llm = OpenAI(model="gpt-3.5-turbo")
    Settings.embed_model = OpenAIEmbedding(model="text-embedding-ada-002")
    Settings.chunk_size = 1024

    # 1. Load documents from a directory
    documents = SimpleDirectoryReader("data").load_data()

    # 2. Create an index from the documents
    index = VectorStoreIndex.from_documents(documents)

    # 3. Create a query engine and query the index
    query_engine = index.as_query_engine()
    response = query_engine.query("What is LlamaIndex?")

    print(f"Query: What is LlamaIndex?")
    print(f"Response: {response.response}")

except Exception as e:
    print(f"An error occurred: {e}")
    print("Please ensure you have `OPENAI_API_KEY` set and `llama-index-llms-openai` and `llama-index-embeddings-openai` installed.")
finally:
    # Clean up dummy file and directory
    if os.path.exists("data/sample_doc.txt"):
        os.remove("data/sample_doc.txt")
    if os.path.exists("data"):
        os.rmdir("data")

view raw JSON →