Biomni
Biomni: a general-purpose biomedical AI agent
What is Biomni?
Biomni: a general-purpose biomedical AI agent
Overview of Biomni
What it is
Biomni is a general-purpose biomedical AI agent designed to autonomously execute a wide range of research tasks across diverse biomedical subfields. It integrates cutting-edge large language model (LLM) reasoning with retrieval-augmented planning and code-based execution to help scientists enhance research productivity and generate testable hypotheses.
Key features
- Autonomously executes biomedical research tasks using natural language instructions - Integrates with multiple LLM providers: Anthropic (Claude), OpenAI, Azure OpenAI, Gemini, Groq, AWS Bedrock, Ollama, and custom models - Retrieval-augmented planning and code-based execution for complex workflows - Includes a pre-built datalake (~11GB) automatically downloaded on first run - Provides a Gradio-based interactive web interface - Supports configuration via .env file or shell environment variables - Open-source with Apache-2.0 license
Use cases
- Plan a CRISPR screen to identify genes that regulate T cell exhaustion and generate 32 genes that maximize the perturbation effect - Perform single-cell RNA-seq annotation and generate meaningful hypotheses - Predict ADMET properties for a given compound - Execute other biomedical research tasks across subfields
Getting started
Install the environment and package: bash # Run the setup script (provided in repository) # Activate the environment conda activate biomni_e1 # Install the official pip package pip install biomni --upgrade # Or install from GitHub source pip install git+https://github.com/snap-stanford/Biomni.git@main Configure API keys using a .env file or shell environment variables (see repository README for details). Basic usage in Python: python from biomni.agent import A1 agent = A1(path='./data', llm='claude-sonnet-4-20250514') agent.go('Plan a CRISPR screen to identify genes that regulate T cell exhaustion')
License and pricing
Licensed under Apache-2.0. Open source, no pricing information stated.
Based on the project's public README and repository metadata.