Dify
by LangGenius
Open-source platform for building LLM apps and agentic workflows
Last reviewed 2026-06-18
Dify is an open-source platform for developing LLM applications and agentic workflows. On a visual canvas you build AI workflows, RAG pipelines, and autonomous agents that use function calling and 50+ built-in tools, with model management, prompt tooling, and observability built in. It integrates hundreds of proprietary and open-source models from many inference providers, including any OpenAI-API-compatible endpoint, and can publish apps as chat UIs or APIs. Dify targets teams that want to move from prototype to production with a self-hostable stack rather than gluing libraries together. It is self-hostable (Docker, Kubernetes via Helm) and also offers Dify Cloud. As a building platform, the autonomy of any app you create is configured by you.
What it can do
Build AI workflows on a visual canvas
SupervisedDesign and test multi-step AI workflows visually, combining prompts, models, logic, and tools.
sourceBuild agents with tools
SupervisedBuild autonomous agents that use function calling and 50+ built-in tools to complete tasks.
sourceRun RAG pipelines
AssistantProvides RAG capabilities from document ingestion (PDF, PPT, and more) through retrieval to ground responses in knowledge bases.
sourceManage models and observability
AssistantIntegrates hundreds of models across providers (including OpenAI-API-compatible endpoints) and monitors logs and performance over time.
source
Strengths
- +Open source and self-hostable (Docker, Kubernetes) with no vendor lock-in
- +Combines workflows, RAG, agents, model management, and observability in one stack
- +Broad model support including any OpenAI-API-compatible endpoint
Limitations
- −A building platform: you design and operate the apps and their guardrails
- −Self-hosting at scale requires infrastructure expertise
- −Agent autonomy is only as good as the workflow you configure
Overview
Dify is an open-source platform for building LLM applications and agentic workflows, combining workflow design, RAG, agents, model management, and observability in one self-hostable stack.
What it does
On a visual canvas you build AI workflows, RAG pipelines, and agents that use function calling and 50+ built-in tools. It integrates hundreds of models across providers (including any OpenAI-API-compatible endpoint), provides prompt tooling, and monitors logs and performance. Apps can be published as chat UIs or APIs.
Classification note
Dify is a building platform, not an end-user agent. The autonomy of any app you create is configured by you, so we list it as supervised-agent conservatively.
Integrations & setup
Self-host via Docker or Kubernetes (Helm charts), or use Dify Cloud. Broad model and provider support, with MCP and function calling.
Pricing
Self-hosting is free and open source. Dify Cloud offers a free sandbox plan plus paid tiers.
Best for / not for
Best for teams that want an open, self-hostable stack to ship LLM apps and agents to production. Less suited to those wanting a fully managed, no-ops product or guaranteed autonomous behavior out of the box.
Alternatives
Flowise and Langflow are open-source visual agent/LLM builders; Stack AI targets enterprise no-code; n8n is automation-first with agent features.
What people are saying
We aggregate real LinkedIn discussion into sentiment for the agents people search most. Dify isn't tracked yet, want it added? Request tracking.
FAQ
Can I self-host Dify?+
Yes. Dify is open source and self-hostable via Docker and on Kubernetes with community Helm charts, and also offers a hosted Dify Cloud with a free sandbox plan.
Is Dify a framework or a platform?+
It is a platform: you build LLM apps, RAG pipelines, and agents on a visual canvas with model management and observability, rather than coding against a library.
Sources
- Dify (official site) · accessed 2026-06-18
- langgenius/dify (GitHub) · accessed 2026-06-18
- Dify introduction (docs) · accessed 2026-06-18
Last reviewed 2026-06-18