Skip to main content


Setup Virtual Environment

When not using a docker container, we recommend using a virtual environment to install AutoGen. This will ensure that the dependencies for AutoGen are isolated from the rest of your system.

Option 1: venv

You can create a virtual environment with venv as below:

python3 -m venv pyautogen
source pyautogen/bin/activate

The following command will deactivate the current venv environment:


Option 2: conda

Another option is with Conda, Conda works better at solving dependency conflicts than pip. You can install it by following this doc, and then create a virtual environment as below:

conda create -n pyautogen python=3.10  # python 3.10 is recommended as it's stable and not too old
conda activate pyautogen

The following command will deactivate the current conda environment:

conda deactivate

Now, you're ready to install AutoGen in the virtual environment you've just created.


AutoGen requires Python version >= 3.8, < 3.12. It can be installed from pip:

pip install pyautogen

pyautogen<0.2 requires openai<1. Starting from pyautogen v0.2, openai>=1 is required.

Migration guide to v0.2

openai v1 is a total rewrite of the library with many breaking changes. For example, the inference requires instantiating a client, instead of using a global class method. Therefore, some changes are required for users of pyautogen<0.2.

  • api_base -> base_url, request_timeout -> timeout in llm_config and config_list. max_retry_period and retry_wait_time are deprecated. max_retries can be set for each client.
  • MathChat is unsupported until it is tested in future release.
  • autogen.Completion and autogen.ChatCompletion are deprecated. The essential functionalities are moved to autogen.OpenAIWrapper:
from autogen import OpenAIWrapper
client = OpenAIWrapper(config_list=config_list)
response = client.create(messages=[{"role": "user", "content": "2+2="}])
  • Inference parameter tuning and inference logging features are currently unavailable in OpenAIWrapper. Logging will be added in a future release. Inference parameter tuning can be done via flaml.tune.
  • seed in autogen is renamed into cache_seed to accommodate the newly added seed param in openai chat completion api. use_cache is removed as a kwarg in OpenAIWrapper.create() for being automatically decided by cache_seed: int | None. The difference between autogen's cache_seed and openai's seed is that:
    • autogen uses local disk cache to guarantee the exactly same output is produced for the same input and when cache is hit, no openai api call will be made.
    • openai's seed is a best-effort deterministic sampling with no guarantee of determinism. When using openai's seed with cache_seed set to None, even for the same input, an openai api call will be made and there is no guarantee for getting exactly the same output.

Optional Dependencies

For the best user experience and seamless code execution, we highly recommend using Docker with AutoGen. Docker is a containerization platform that simplifies the setup and execution of your code. Developing in a docker container, such as GitHub Codespace, also makes the development convenient.

When running AutoGen out of a docker container, to use docker for code execution, you also need to install the python package docker:

pip install docker

pyautogen<0.2 offers a cost-effective hyperparameter optimization technique EcoOptiGen for tuning Large Language Models. Please install with the [blendsearch] option to use it.

pip install "pyautogen[blendsearch]<0.2"

Example notebooks:

Optimize for Code Generation

Optimize for Math

pyautogen supports retrieval-augmented generation tasks such as question answering and code generation with RAG agents. Please install with the [retrievechat] option to use it.

pip install "pyautogen[retrievechat]"

RetrieveChat can handle various types of documents. By default, it can process plain text and PDF files, including formats such as 'txt', 'json', 'csv', 'tsv', 'md', 'html', 'htm', 'rtf', 'rst', 'jsonl', 'log', 'xml', 'yaml', 'yml' and 'pdf'. If you install unstructured (pip install "unstructured[all-docs]"), additional document types such as 'docx', 'doc', 'odt', 'pptx', 'ppt', 'xlsx', 'eml', 'msg', 'epub' will also be supported.

You can find a list of all supported document types by using autogen.retrieve_utils.TEXT_FORMATS.

Example notebooks:

Automated Code Generation and Question Answering with Retrieval Augmented Agents

Group Chat with Retrieval Augmented Generation (with 5 group member agents and 1 manager agent)

Automated Code Generation and Question Answering with Qdrant based Retrieval Augmented Agents

To use TeachableAgent, please install AutoGen with the [teachable] option.

pip install "pyautogen[teachable]"

Example notebook: Chatting with TeachableAgent

  • Large Multimodal Model (LMM) Agents

We offered Multimodal Conversable Agent and LLaVA Agent. Please install with the [lmm] option to use it.

pip install "pyautogen[lmm]"

Example notebooks:

LLaVA Agent

pyautogen<0.2 offers an experimental agent for math problem solving. Please install with the [mathchat] option to use it.

pip install "pyautogen[mathchat]<0.2"

Example notebooks:

Using MathChat to Solve Math Problems