Capability · Framework — orchestration

Dify

Dify is an opinionated platform for shipping LLM apps: it bundles a workflow canvas, prompt IDE, RAG with document ingestion, agent tool-use, and a polished end-user chat UI. Teams treat Dify less as a library and more as a hosted backend — you ship a URL or SDK to your app, not a framework to maintain.

Framework facts

Category: orchestration
Language: Python / TypeScript
License: Apache-2.0 (with some restrictions)
Repository: https://github.com/langgenius/dify

Install

git clone https://github.com/langgenius/dify.git
cd dify/docker
cp .env.example .env
docker compose up -d
# open http://localhost

Quickstart

# After install, create an app in the UI, copy the API key, then:
curl -X POST 'http://localhost/v1/chat-messages' \
  -H 'Authorization: Bearer app-XXXX' \
  -H 'Content-Type: application/json' \
  -d '{"query":"hi","user":"abc","response_mode":"blocking","inputs":{}}'

Alternatives

Flowise / Langflow — lighter visual builders without BaaS UI
BISHENG — Chinese open-source peer
OpenWebUI — chat UI first, less workflow-focused
LangChain + LangGraph Studio — code-first alternative

Frequently asked questions

Is Dify free for commercial use?

The core is open-source under a modified Apache-2.0 with additional commercial restrictions (notably you can't resell it as a competing multi-tenant service). Always re-check the LICENSE at the time you fork.

Can Dify replace my RAG backend?

For many SMB use cases, yes — Dify handles ingestion, chunking, retrieval, and chat. For tight control over embedding pipelines, reranking, or custom eval you may outgrow it and move to LangChain/LlamaIndex.

Sources

Dify — docs — accessed 2026-04-20
Dify GitHub — accessed 2026-04-20