Dify

Dify is a platform for building LLM apps with visual workflows, RAG, and agent tools-usable in Dify Cloud or self-hosted for more control.

LLM Application PlatformSource-available under an Apache 2.0–based license with additional restrictionsCloud + Self-hosted

Overview

What is Dify?

Dify is a platform for developing LLM applications, designed to help you move from prototype to production using a UI instead of building everything from scratch.

It supports building workflows that include AI agents and tool-using logic, plus RAG pipelines, plus model management and logging and monitoring features to help you run and improve apps over time.

Dify can be used through Dify Cloud (hosted) or deployed in your own environment via self-hosting, depending on how much control you need.

Key Features

Visual workflow builder: Build and test AI workflows on a visual canvas
RAG pipeline support: RAG features covering ingestion → retrieval, including common document formats
Agent capabilities: Define agents (e.g., using function calling or ReAct) and give them tools
Model/provider support: Integrates with many proprietary/open-source LLMs and OpenAI API–compatible models
APIs / Backend-as-a-Service: Dify provides APIs so you can integrate apps into your own systems

Common Use Cases

•Build chat or workflow-based LLM applications with UI + API access
•Create RAG apps that answer questions from internal documents/knowledge bases
•Build AI agent workflows where an agent uses tools as part of a repeatable process

Who It's For

Dify fits teams and builders who want a platform (UI + APIs) to develop and operate LLM apps-especially when they need workflows/RAG/agent tools and the option to self-host.

How It Works

Main Approach

You build apps and workflows in Dify's interface (including a visual workflow canvas) and ship them as usable applications and APIs.

LLM/Model Integration

Dify connects to multiple LLM providers (including OpenAI API–compatible options) so you can choose and switch models for your app.

Application Building

Dify is designed for building LLM applications (not just single prompts), with workflows, RAG, and agent tooling as building blocks.

Key Differentiators

✓Platform for LLM apps (UI + APIs), not only code libraries
✓Combines workflows + RAG + agent capabilities in one place
✓Cloud or self-hosted deployment options
✓License is Apache 2.0–based with additional restrictions (important for commercial/multi-tenant use)

Pricing & Plans

Dify's plans and pricing can change, so the safest approach is to reference the official pricing page directly.

View official pricing

Getting Started

Read the official Dify documentation

Quick Start

1Choose Dify Cloud (hosted) or self-hosting depending on your needs
2Create an app/workflow in the Dify UI (workflow canvas / agent tools as needed)
3Test, then integrate using Dify's APIs (or deploy your self-hosted instance)

Try Dify

When to Use Dify

Dify vs LangChain

Dify is a platform with UI + deployment and operations features, while LangChain is a code-first framework/library approach. (Different building styles: platform vs framework.)

Learn more about LangChain

Dify vs Relevance AI

Dify is positioned around building LLM applications (workflows/RAG/agents + APIs), while Relevance AI is typically discussed as an agent-focused platform. Use both for AI agent workflows, but they're optimized for different build styles.

Learn more about Relevance AI

Related Resources

Back to AI Agent Builders

Related Tools

n8n Make

Last updated: December 2025

LLM Application Platform