Repolode
Submit
camel-ai

camel-ai/oasis

A Python simulator where LLM agents mimic up to one million users on Twitter- and Reddit-like social platforms to study information spread and polarization.

View on GitHubWebsite
Stars
5,317
Forks
653
Open issues
59
Last commit
yesterday

What is it?

OASIS is an open-source social media simulator built on large language model agents. It models user behavior on platforms resembling Twitter and Reddit, with simulations scaling to one million agents. It is aimed at researchers who want to study phenomena such as information spread, group polarization and herd behavior.

Agents can take 23 actions, including following, commenting and reposting. The environment adapts to changes in the social network and content over time, and includes interest-based and hot-score-based recommendation algorithms that determine what users see.

The project is installed from PyPI as camel-oasis and runs on the CAMEL framework. Users define agent profiles from JSON files, choose which actions are available, and step the environment with either scripted or LLM-driven actions, with state stored in a database.

Key features

  • 01Simulates up to one million LLM agents on Twitter- and Reddit-style platforms.
  • 02Supports 23 user actions such as following, commenting and reposting.
  • 03Includes interest-based and hot-score-based recommendation algorithms.
  • 04Dynamically updates social networks and content during a simulation.
  • 05Installable from PyPI as camel-oasis and built on the CAMEL framework.

Use cases

  • Studying how information and misinformation spread through a social network.
  • Researching group polarization and herd behavior with many simulated users.
  • Testing how recommendation algorithms shape user interactions before deployment.
  • Generating synthetic social media interaction data for analysis.

Quick start

  1. 1

    Install the OASIS package from PyPI

    pip install camel-oasis
  2. 2

    Set your OpenAI API key in the shell

    export OPENAI_API_KEY=<insert your OpenAI API key>
  3. 3

    Download the example profile file user_data_36.json into ./data/reddit

  4. 4

    Run the example Python code from the README

Things to know

  • Requires a paid OpenAI API key (or another configured model provider) to run LLM agents.
  • Token costs grow with the number of agents, activation probability and time steps.

Topics

#agent-based-simulation#llm-agents#multi-agent-systems#social-media#large-language-models#python#agent-based-framework#ai-societies#deep-learning#large-scale

FAQ

What is oasis?
A Python simulator where LLM agents mimic up to one million users on Twitter- and Reddit-like social platforms to study information spread and polarization. OASIS is an open-source social media simulator built on large language model agents. It models user behavior on platforms resembling Twitter and Reddit, with simulations scaling to one million agents. It is aimed at researchers who want to study phenomena such as information spread, group polarization and herd behavior.
Is oasis free to use? What license does it have?
oasis is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is oasis still maintained?
Actively maintained. The last commit to oasis was on Oct 19, 2026.
What can I use oasis for?
Studying how information and misinformation spread through a social network. Researching group polarization and herd behavior with many simulated users. Testing how recommendation algorithms shape user interactions before deployment. Generating synthetic social media interaction data for analysis.
How do I install oasis?
Install the OASIS package from PyPI: pip install camel-oasis Set your OpenAI API key in the shell: export OPENAI_API_KEY=<insert your OpenAI API key> Download the example profile file user_data_36.json into ./data/reddit Run the example Python code from the README
Where can I get oasis?
The source code and installation instructions are on GitHub: https://github.com/camel-ai/oasis

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects