camel
An open-source multi-agent framework for building and scaling AI agent societies.
Explore scaling laws and build large-scale multi-agent systems with stateful memory.
CAMEL is an open-source framework designed for studying multi-agent systems and scaling laws. It allows you to simulate up to one million communicating agents with stateful memory and tool integration.
What it is
The framework implements various agent roles, tasks, prompts, and simulated environments to automate data generation and task execution. It uses a Code-as-Prompt design principle where code and comments serve as explicit instructions for agents.
Who it's for
It suits AI researchers and developers studying multi-agent interactions, synthetic data generation, and agent scaling laws. It does not suit those looking for a simple plug-and-play single chatbot application.
Key features
Large-Scale Simulation
Simulate up to one million agents to study emergent behaviors.
Stateful Memory
Agents retain and leverage historical context for multi-step interactions.
Dynamic Communication
Real-time interactions among agents foster seamless task collaboration.
Tool Integration
Integrate search toolkits and external APIs directly into chat agents.
Up and running in 3 steps
Our honest verdict
What's great
- Supports simulation of up to one million agents for complex emergent behavior studies.
- Built-in stateful memory handling for multi-step environment interactions.
- Includes specialized search toolkits and web tools for agent capabilities.
- Comprehensive logging system for model requests, responses, and configurations.
Where it falls short
- Requires active API keys and ongoing model usage costs, such as OpenAI GPT-4o.
- Setting up large-scale agent societies demands deep technical understanding of LLMs.
- Managing environment configurations and log directories requires manual configuration.
Pricing & license
The framework is free under its open-source license, but running agents requires API access to underlying models like OpenAI, which incur separate usage costs.
Open remote AI-training roles
880 open remote AI-training roles, updated November 15, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
These reviewed AI tools do the same job instantly — from our scored directory.
Use it, or have us set it up
FAQ
What is the primary goal of CAMEL?
CAMEL is dedicated to finding the scaling laws of agents by studying multi-agent behaviors, capabilities, and risks on a large scale.
How do I enable model request logging?
You can enable logging by setting environment variables such as CAMEL_MODEL_LOG_ENABLED=true and specifying a CAMEL_LOG_DIR.
