AI Agent GitHub Stars Ranking 2026

Data Notes

The following data comes from the GitHub API, collected in July 2026.

Overall Top 10

1. n8n-io/n8n — 196k+ Stars github.com/n8n-io/n8n

n8n was originally built for workflow automation — similar to Zapier, but open-source and self-hostable. After adding AI Agent features in 2023, it absolutely took off. Its strength lies in visual drag-and-drop — you can build an AI workflow without writing code, chaining together LLM calls, data transformations, and API integrations. It currently supports 300+ integration nodes, from Slack to databases to various AI models. It's great for business users who want to quickly deploy AI workflows, and equally useful for developers doing prototype validation. From having almost no connection to the Agent ecosystem to topping the Agent charts at 196k Stars, n8n is the biggest dark horse of the past two years.

2. Significant-Gravitas/AutoGPT — 185k+ Stars github.com/Significant-Gravitas/AutoGPT

The phenomenon project of spring 2023. You give it a goal, and it breaks it down into subtasks, calls tools, and iterates until completion. At the time, everyone was blown away by the concept — AI was no longer just passively answering questions, but actively getting things done. Although AutoGPT's technical approach now looks somewhat primitive compared to mature Agent frameworks (single-threaded, prompt-loop based), its historical significance is irreplaceable. 185k Stars is proof that it once ignited an entire era.

3. ollama/ollama — 176k+ Stars github.com/ollama/ollama

The de facto standard for running large models locally. In one sentence: if you want to run Llama, Mistral, Qwen, or similar models on your own machine, ollama is the simplest way. One command to download a model, one command to start the service, and then any OpenAI-compatible client can connect to it. It doesn't do Agents, and it doesn't do UIs — it's the infrastructure layer of the Agent ecosystem. Without ollama, local Agent adoption wouldn't have spread this fast. The model library currently supports over 100 models.

4. langgenius/dify — 148k+ Stars github.com/langgenius/dify

An AI application development platform built by a Chinese team with excellent internationalization. Dify turns prompt engineering, RAG (Retrieval-Augmented Generation), Agent orchestration, and model management into a visual platform. You can build an AI assistant with a knowledge base without writing a single line of code, or go deep with custom code. From 2025 to 2026, Stars doubled from 74k to 148k — an incredibly strong growth trajectory. It's now arguably the go-to application development platform in the Chinese AI community.

5. open-webui/open-webui — 145k+ Stars github.com/open-webui/open-webui

A beautiful LLM chat interface, originally designed for Ollama. But it later added support for OpenAI-compatible APIs, so you can use it to connect to any large model. Its UI is better than many commercial products — conversation management, prompt templates, RAG, multi-user support, everything you'd expect is there. Gaining 100k+ Stars in a year shows how strong the demand is for a good interface among local AI users. A lot of people run this setup: ollama for models + open-webui as the frontend = your own private ChatGPT.

6. langchain-ai/langchain — 141k+ Stars github.com/langchain-ai/langchain

One of the earliest Agent development frameworks, and the one with the richest ecosystem. It provides a complete toolchain — prompt templates, chained calls, Agent frameworks, document loaders, and vector store integrations. Almost any large-model development need has a corresponding LangChain module you can plug in. But its growth curve has plateaued — not because it's bad, but because higher-level, more abstract tools (Dify, n8n) have taken away some users. If you need to write code for deep customization, LangChain is still the safest choice.

7. ggml-org/llama.cpp — 120k+ Stars github.com/ggml-org/llama.cpp

The low-level infrastructure for LLM inference engines. Implemented in pure C++ with no Python dependencies, it runs on virtually any hardware — from servers to laptops to Raspberry Pis. It turned large-model inference from a cloud-only privilege into something anyone can do. Many higher-level tools (including ollama) call llama.cpp under the hood. Without it, the local AI ecosystem wouldn't be nearly as thriving as it is today. 120k Stars is a confirmation of its value.

8. nomic-ai/gpt4all — 77k+ Stars github.com/nomic-ai/gpt4all

A local LLM ecosystem tool focused on ease of use. Download, install, run — three steps and you're done. It comes with a polished chat interface built in, and its model library covers models of all sizes and purposes. Compared to ollama, gpt4all is more consumer-oriented — you don't need to know the command line to use it. Its growth tells the same story: everyday people are also trying to bring AI onto their own computers.

9. microsoft/autogen — 59k+ Stars github.com/microsoft/autogen

A Microsoft product — a multi-Agent conversation framework. The core idea is to have multiple AI Agents work like a group of people in a meeting, dividing labor and collaborating to complete complex tasks. For example, one Agent writes code, one tests it, one reviews it, and they iterate with feedback. In enterprise Agent application scenarios, Autogen is a heavyweight choice. Microsoft itself uses it to build quite a few internal tools.

10. crewAIInc/crewAI — 55k+ Stars github.com/crewAIInc/crewAI

A multi-Agent collaboration framework, similar to Autogen but more lightweight. You define Roles, Tasks, and Processes, then let Agents operate like a team. Its API design is intuitive, and Python developers can get up to speed very quickly. Growing from roughly 20k Stars in 2025 to 55k+ — a 2.5x increase — shows that multi-Agent collaboration is becoming a mainstream requirement.

Fastest-Growing Projects

Key Observations

1. n8n is the biggest dark horse — transformed from an automation tool into an Agent orchestration platform, topping the charts at 196k
2. Dify grew the fastest — doubled from 74k to 148k; visual platforms for RAG + Agent are becoming mainstream
3. LangChain's growth has slowed — Agent development is becoming more tool-ified, and higher-level tools are eating into its user base
4. ollama + open-webui are growing symbiotically — 321k+ Stars combined; the local AI ecosystem has truly arrived

Framework Selection Guide

• Want visual building → n8n (workflows) / Dify (RAG + Agent)
• Want code-level control → LangChain / CrewAI
• Want local deployment → ollama + open-webui
• Want to study Agent fundamentals → AutoGPT / Autogen

Summary

GitHub Stars reflect community momentum and ecosystem maturity. When choosing a framework, Stars are an important reference but not the only criterion — what fits your needs is what's best.

💡 What You Will Learn

## Data Notes The following data comes from the GitHub API, collected in July 2026. ### Overall Top 10 **1. n8n-io/n8n — 196k+ Stars** <a href="https://github.com/n8n-io/n8n" target="_blank">github

📜 Table of Contents

• Data Notes
• Overall Top 10
• Fastest Growing Projects
• Key Observations
• Framework Selection Advice
• Summary

Data Notes

The following data comes from the GitHub API, collected in July 2026.

Overall Top 10

1. n8n-io/n8n — ⭐ 196k+ github.com/n8n-io/n8n

n8n was originally a workflow automation tool, similar to Zapier, but open-source and self-hostable. After adding AI Agent features in 2023, it took off. Its strength lies in visual drag-and-drop—you can build an AI workflow without writing code, connecting LLM calls, data transformations, and API integrations. It currently supports over 300 integration nodes, from Slack to databases to various AI models. It's perfect for business users to quickly implement AI workflows, and for developers to prototype. From having almost no connection to the Agent ecosystem to topping the Agent charts with 196k Stars, n8n is the biggest dark horse of the past two years.

2. Significant-Gravitas/AutoGPT — ⭐ 185k+ github.com/Significant-Gravitas/AutoGPT

A phenomenon project from spring 2023. Give AI a goal, and it can break it down into sub-tasks, call tools, and iterate until completion. At the time, everyone was blown away by the concept—AI was no longer passively answering questions, but actively completing tasks. Although AutoGPT's technical approach now seems somewhat primitive (single-threaded, prompt-loop-based) as Agent frameworks matured, its historical significance is irreplaceable. 185k Stars prove it once ignited an era.

3. ollama/ollama — ⭐ 176k+ github.com/ollama/ollama

The de facto standard for running large models locally. In a nutshell: ollama is the simplest way to run models like Llama, Mistral, and Qwen locally. One command to download a model, one command to start the service, and then any OpenAI-compatible client can connect to it. It doesn't do Agents or interfaces—it's the infrastructure layer for the Agent ecosystem. Without ollama, local Agent adoption wouldn't have spread this fast. It currently supports a model library of over 100 models.

4. langgenius/dify — ⭐ 148k+ github.com/langgenius/dify

An AI application development platform, built by a Chinese team with excellent internationalization. Dify turns Prompt engineering, RAG (Retrieval-Augmented Generation), Agent orchestration, and model management into a visual platform. You can build an AI assistant with a knowledge base without writing a single line of code, or dive deep with custom code. From 2025 to 2026, Stars doubled from 74k to 148k, showing very strong growth. It's now the go-to application development platform in the Chinese AI community.

5. open-webui/open-webui — ⭐ 145k+ github.com/open-webui/open-webui

A beautiful LLM chat interface, originally designed for Ollama. But it later added support for OpenAI-compatible APIs, so you can use it to connect to any large model. Its UI is better than many commercial products—session management, Prompt templates, RAG, multi-user support, everything you need. Gaining 100k+ Stars in a year shows the strong demand for a good interface among local AI users. Many people set it up like this: ollama runs the model + open-webui as the frontend = your own private ChatGPT.

6. langchain-ai/langchain — ⭐ 141k+ github.com/langchain-ai/langchain

One of the earliest Agent development frameworks, and the one with the richest ecosystem. It provides a complete toolchain—Prompt templates, chain calls, Agent frameworks, document loaders, vector store integrations. Almost any development need related to large models has a corresponding module in LangChain. But its growth curve has plateaued, not because it's bad, but because higher-level, more abstract tools (Dify, n8n) have taken some users. For deep customization with code, LangChain is still the safest bet.

7. ggml-org/llama.cpp — ⭐ 120k+ github.com/ggml-org/llama.cpp

The underlying infrastructure for LLM inference engines. Implemented in pure C++ with no Python dependencies, it runs on various hardware—from servers to laptops to Raspberry Pis. It turned large model inference from a cloud privilege into something anyone can do. Many higher-level tools (including ollama) use llama.cpp under the hood. Without it, the local AI ecosystem wouldn't be as thriving as it is today. 120k Stars confirm its value.

8. nomic-ai/gpt4all — ⭐ 77k+ github.com/nomic-ai/gpt4all

A local LLM ecosystem tool focused on ease of use. Download, install, run—three steps. It has a built-in, user-friendly chat interface, and its model library covers various sizes and use cases. Compared to ollama, gpt4all is more consumer-oriented—you can use it without knowing the command line. Its growth highlights the same trend: ordinary people are also trying to bring AI to their own computers.

9. microsoft/autogen — ⭐ 59k+ github.com/microsoft/autogen

A multi-Agent conversation framework from Microsoft. The core idea is to have multiple AI Agents discuss like a group of people, dividing and collaborating to complete complex tasks. For example, one Agent writes code, one tests it, one reviews it, and they iterate with feedback. In enterprise-level Agent application scenarios, Autogen is a heavyweight choice. Microsoft itself uses it for many internal tools.

10. crewAIInc/crewAI — ⭐ 55k+ github.com/crewAIInc/crewAI

A multi-Agent collaboration framework, similar to Autogen but lighter. Define Roles, Tasks, and Processes, then let Agents operate like a team. Its API design is intuitive, making it very easy for Python developers to get started. Growing from about 20k Stars in 2025 to 55k+, a 2.5x increase, shows that multi-Agent collaboration is becoming a mainstream need.

Fastest Growing Projects

Key Observations

1. n8n is the biggest dark horse — Transitioned from an automation tool to an Agent orchestration platform, topping the charts at 196k
2. Dify grew the fastest — 74k→148k doubled, visual platforms for RAG+Agent are becoming mainstream
3. LangChain growth slowed — Agent development is becoming tool-ified, with higher-level tools taking users
4. ollama+open-webui symbiotic growth — Combined 321k+ Stars, the local AI ecosystem is well-established

Framework Selection Advice

• Want visual building → n8n (workflows) / Dify (RAG+Agent)
• Want code control → LangChain / CrewAI
• Want local operation → ollama + open-webui
• Want to study Agent principles → AutoGPT / Autogen

Summary

GitHub Stars reflect community vitality and ecosystem maturity. Stars are an important reference when choosing a framework, but not the only criterion—the best fit is what matters.

📎 Related Recommendations: - 2026 Python AI Agent Framework Comparison: LangChain to CrewAI, Who's Hottest? - 2026 Local AI Agent Setup Tutorial: From Scratch