#9393rd of 102 in Agent platformsFor science, health and research, science, health and medicine

biomni

Biomedical research agent that plans and runs code across bio tools and databases

Stars
4.1k
License
Apache-2.0
Last commit
Nov 2026
Last release
Oct 2025
Language
Python
Stars, last 29 weeks+1,104
25 Apr 2026: 2,988 stars2 May 2026: 3,006 stars9 May 2026: 3,030 stars16 May 2026: 3,060 stars23 May 2026: 3,087 stars30 May 2026: 3,111 stars6 Jun 2026: 3,140 stars13 Jun 2026: 3,178 stars20 Jun 2026: 3,211 stars27 Jun 2026: 3,240 stars4 Jul 2026: 3,275 stars11 Jul 2026: 3,350 stars18 Jul 2026: 3,488 stars25 Jul 2026: 3,538 stars1 Aug 2026: 3,598 stars8 Aug 2026: 3,651 stars15 Aug 2026: 3,695 stars22 Aug 2026: 3,750 stars29 Aug 2026: 3,798 stars5 Sept 2026: 3,832 stars12 Sept 2026: 3,870 stars19 Sept 2026: 3,890 stars26 Sept 2026: 3,910 stars3 Oct 2026: 3,936 stars10 Oct 2026: 3,958 stars17 Oct 2026: 3,987 stars24 Oct 2026: 4,015 stars31 Oct 2026: 4,042 stars7 Nov 2026: 4,074 stars14 Nov 2026: 4,092 stars

2,988 on 25 Apr 20264,092 on 14 Nov 2026

From GitHub's daily star counts.

Checked on GitHub.

Overview

Biomni is a Python agent (class A1) that takes natural-language biomedical tasks, such as CRISPR screen planning, scRNA-seq annotation or ADMET prediction, and runs LLM-generated code against bundled tools, software and an ~11GB data lake. It works with Anthropic, OpenAI, Azure, Gemini, Bedrock, Groq, Ollama or custom endpoints, supports MCP servers, and includes a Gradio UI.

Who it is for: Biomedical researchers and bioinformaticians automating analysis with LLM agents

Strengths

  • Many LLM providers supported, including Ollama and any OpenAI-compatible custom endpoint
  • Bundles biomedical tools, databases and a retrievable Know-How protocol library
  • Adds external tools via MCP; local Gradio UI served on port 7860
  • Ships Biomni-R0 32B reasoning model and Biomni-Eval1 benchmark (433 instances)

Weaknesses

  • Executes LLM-generated code with full system privileges; needs external sandboxing
  • Large conda environment via setup script plus ~11GB data lake download on first run
  • Release frozen at April 2025, so it differs from the hosted web platform
  • Some integrated tools and databases carry restrictive commercial licenses; some packages conflict

What it needs

  • GPU optional
  • Needs conda, LLM API key (Anthropic by default), ~11GB data lake, Gradio 5.x (optional), SGLang (optional, for Biomni-R0)
  • Models: Anthropic, OpenAI, Azure OpenAI, Gemini, AWS Bedrock
  • port 7860

What changed

From the release notes. No flag means the note does not say so, not that the upgrade is safe.

Also in Agent platforms

See all 102
Also in Agent platforms
RankProjectScore
1AutoGPTRole-based AI agents coordinated by one lead agent, plus a visual workflow builder Live demo ↗ (opens in a new tab)188.1k starscustom license95 out of 100
2daily_stock_analysisLLM-driven daily analysis of watchlist stocks across six markets, pushed to chat apps73.1k starsMIT95 out of 100
3LangflowVisual builder for AI agents and workflows, served as API or MCP server156.7k starsMIT94 out of 100
#4deer-flowAgent harness running sub-agents, memory and sandboxes with pluggable skills85.5k starsMIT93 out of 100
#5DifyVisual builder for LLM workflows, RAG pipelines and agents, with APIs Live demo ↗ (opens in a new tab)161.4k starscustom license92 out of 100