Dify
Build and operate AI agents, visual workflows, and RAG pipelines on an open-source development platform.
Design AI workflows, manage model providers, and deploy RAG pipelines without starting from scratch.
Dify is an open-source LLM application development platform featuring a visual workflow canvas and agent builder. It helps software teams move from rapid prototyping to production with built-in observability and backend APIs. You should use it if you want to deploy custom AI agents and document pipelines without building infrastructure from zero.
What it is
Dify connects dozens of model providers with document ingestion, function calling, and prompt engineering tools. It provides a visual canvas to link LLMs, 50+ built-in agent tools, and observability integrations like Langfuse and Phoenix. The platform exposes full Backend-as-a-Service APIs to connect your AI logic directly into your app backend.
Who it's for
It suits software engineers and technical product teams who need to build, test, and host complex LLM workflows or RAG applications. It is not ideal for non-technical users who cannot manage Docker environments or server infrastructure. Teams looking for purely offline desktop utilities will also find it unnecessary.
Key features
Visual Canvas Workflow Builder
Create and test complex AI application logic on an interactive node-based canvas.
Broad Model Provider Support
Connect proprietary and open-source models across dozens of providers and local OpenAI-compatible APIs.
Built-In RAG Pipeline
Extract and index document formats including PDFs and PPTs for immediate vector retrieval.
Agent Tooling Framework
Configure agents using Function Calling or ReAct with tools like Google Search and WolframAlpha.
Backend as a Service
Access corresponding API endpoints for every visual workflow to integrate AI features into your code.
Up and running in 3 steps
Our honest verdict
What's great
- Includes over 50 pre-built agent tools like Google Search, DALL-E, and WolframAlpha out of the box.
- Exposes Backend-as-a-Service APIs for every created workflow to simplify backend integration.
- Integrates directly with third-party observability platforms like Opik, Langfuse, and Arize Phoenix.
- Supports document parsing for common formats like PDF and PPT directly inside the RAG pipeline.
Where it falls short
- Requires hosting infrastructure with at least 2 CPU cores and 4 GiB RAM to run properly.
- Running high-throughput RAG indexing and embedding storage increases ongoing server hosting costs.
- Self-managed deployments require technical maintenance of Docker containers and environment variables.
Pricing & license
Dify is free to self-host under its open-source license. The hosted Dify Cloud service offers a free sandbox plan with 200 free GPT-4 calls before requiring a paid plan. Self-hosted deployments require you to pay for your own server hosting and third-party API model usage.
Open remote AI-training roles
876 open remote AI-training roles, updated November 15, 2026; most pay $20–$170 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
Use it, or have us set it up
FAQ
What hardware is required to self-host Dify?
You need a machine with at least 2 CPU cores and 4 GiB of RAM, along with Docker and Docker Compose v2.24.0 or later.
Does Dify support self-hosted LLM models?
Yes, it supports any OpenAI API-compatible model alongside dozens of major cloud inference providers.
Can I connect Dify to external monitoring systems?
Yes, Dify natively integrates with observability platforms such as Opik, Langfuse, and Arize Phoenix.
