Rd Agent
Rd Agent is Research and development (R&D) is crucial for the enhancement of industrial productivity, especially in the AI era, where the core aspects of R&D are mainly focused on data and models. We are committed to automating these high-value generic R&D processes through R&D-Agent, which lets AI drive data-d
Last verified:
What is Rd Agent?
Rd Agent is RD-Agent is an open-source research and development automation tool created by Microsoft that automates the most critical and valuable aspects of industrial R&D processes, particularly in data-driven AI scenarios. The tool focuses on streamlining the development of models and data through a framework with two key components: 'R' for proposing new ideas through research, and 'D' for implementing them through development. It lets AI drive data-driven AI by automating high-value generic R&D processes.
Rd Agent pricing
Pricing model: Freemium
RD-Agent is completely free and open-source under the MIT license. There is no paid tier or subscription model. Users only need to pay for the LLM API calls they use (OpenAI, Azure OpenAI, DeepSeek, or other LiteLLM providers). Experiments can run at very low cost, with quant trading experiments demonstrated under $10. The tool itself has no licensing fees, and users can install it directly from PyPI with 'pip install rdagent' or from source.
Rd Agent pros
- Top-performing ML engineering agent on MLE-bench benchmark
- Open-source with MIT license and 6.9k GitHub stars
- Automates full-stack quantitative strategy development
- Achieves 2× higher ARR than benchmark factor libraries
- Uses 70% fewer factors than classical libraries
- Cost-efficient with experiments under $10
- Supports multiple LLM providers via LiteLLM backend
- Compatible with DeepSeek models for cost-effective inference
- Multi-agent framework with coordinated factor-model co-optimization
- Iterative self-loop for continuous improvement from feedback
- Auto-reads research papers and financial reports
- Automated feature engineering for Kaggle competitions
- Supports Azure OpenAI and OpenAI API directly
- Health check command for easy environment validation
- Active community with Discord and WeChat groups
- Multiple ready-to-use scenarios (finance, medical, general)
- Knowledge forest for continuously refining hypotheses
Rd Agent cons
- Only supports Linux operating system
- Requires Docker installation before use
- Requires Conda environment with Python 3.10 or 3.11
- Needs API keys for LLM providers (OpenAI, Azure, DeepSeek)
- Configuration can be complex with multiple environment variables
- Port 19899 must be available for UI monitoring
- Not ready-to-use for financial investment without testing
- Challenges adapting to volatile market regimes in quant trading
- Embedding model requirements vary by provider
- Reasoning models require additional REASONING_THINK_RM setting
Frequently asked questions about Rd Agent
What is RD-Agent?
RD-Agent is an open-source research and development automation tool by Microsoft that automates the most critical aspects of industrial R&D processes, especially in data-driven AI scenarios. It uses a framework with 'R' for proposing new ideas through research and 'D' for implementing them through development, enabling AI to drive data-driven AI by automating high-value generic R&D processes.
Is RD-Agent free to use?
Yes, RD-Agent is completely free and open-source under the MIT license. You can install it directly from PyPI with 'pip install rdagent'. The only costs involved are the LLM API calls you make through providers like OpenAI, Azure OpenAI, or DeepSeek, which can be as low as under $10 for experiments.
What operating systems does RD-Agent support?
RD-Agent currently only supports Linux operating systems. Users must ensure they are running Linux before attempting to use most scenarios. This is a current limitation of the tool.
What are the system requirements for RD-Agent?
RD-Agent requires Docker installed (with the user able to run Docker commands without sudo), a Conda environment with Python 3.10 or 3.11, and API access to an LLM provider supporting ChatCompletion, json_mode, and embedding query. Port 19899 should be available for the UI monitoring feature.
What scenarios does RD-Agent support?
RD-Agent supports multiple scenarios including: quantitative finance (iterative factor/model evolution, factor extraction from financial reports), medical (prediction model evolution), general data science (paper reading and implementation), and Kaggle competitions (auto model tuning and feature engineering). Each scenario has specific commands like 'rdagent fin_quant', 'rdagent data_science', and 'rdagent general_model'.
How does RD-Agent perform on benchmarks?
RD-Agent currently leads as the top-performing machine learning engineering agent on MLE-bench, a comprehensive benchmark evaluating AI agents on machine learning engineering tasks using 75 Kaggle competitions. R&D-Agent o1-preview achieves 22.4% overall success rate, outperforming AIDE o1-preview at 16.9%.
Which LLM providers does RD-Agent support?
RD-Agent supports multiple LLM providers through LiteLLM backend, including OpenAI (gpt-4o, gpt-4-turbo), Azure OpenAI, and DeepSeek models. It requires ChatCompletion, json_mode, and embedding query capabilities. Users can configure separate API bases for chat and embedding models for maximum flexibility.
How do I install RD-Agent?
For users, install directly from PyPI with 'pip install rdagent'. For developers wanting the latest version, clone the repository with 'git clone https://github.com/microsoft/RD-Agent', cd into it, and run 'make dev'. First, create a Conda environment with 'conda create -n rdagent python=3.10' and ensure Docker is installed.
What is RD-Agent(Q)?
RD-Agent(Q) is the quantitative finance version of RD-Agent, the first data-centric multi-agent framework designed to automate full-stack research and development of quantitative strategies via coordinated factor-model co-optimization. It achieves approximately 2× higher ARR than benchmark factor libraries while using over 70% fewer factors, with experiments running under $10 on CSI 300.
How do I monitor RD-Agent results?
You can monitor application results by running 'rdagent ui --port 19899 --log_dir <your log folder>' to view run logs through a web UI. The 'data_science' parameter should be set to True if you want to see logs for the data science scenario, otherwise False. Check if port 19899 is available using 'rdagent health_check' before running.