Dify

Dify is an open-source platform for building and running LLM applications, RAG pipelines and agentic workflows through a visual builder backed by production infrastructure. It suits enterprise teams that want business and engineering staff to prototype and ship AI apps on cloud, VPC or self-hosted infrastructure without assembling their own orchestration, retrieval and monitoring stack.

Dify is an open-source platform from LangGenius for building, deploying and operating LLM applications and agentic workflows. It pairs a visual Workflow Studio with production plumbing: a prompt IDE, a Knowledge Pipeline for RAG ingestion and retrieval, agent configuration with tools and memory, a marketplace of model-provider and tool plugins, Model Context Protocol support and built-in monitoring, and it publishes finished apps as web apps, APIs or embeds. It connects to hundreds of models, including OpenAI GPT, Mistral and Llama. Dify runs as a managed cloud service, as a self-hosted or VPC Enterprise edition with SSO, SAML and RBAC, or as a free Community Edition deployed with Docker Compose; its GitHub repository has more than 155,000 stars. In March 2026 the Menlo Park company, led by CEO Luyu Zhang, raised a $30M Series Pre-A at a $180M valuation led by HSG, reporting that Dify runs on more than 1.4 million machines, with over 2,000 teams on commercial versions and more than 280 enterprises including Maersk, ETS, Anker Innovations and Novartis. Release 1.16.0 on July 17, 2026 introduced Dify Agent in open beta, an agent builder with a built-in sandbox for code and shell execution and a skill system, and the following 1.17.0 release added E2B cloud sandboxes, workspace-level skill management and provider-neutral tracing. Its licence is Apache 2.0 with added conditions: no multi-tenant hosting without written authorisation and no removal of the console logo. Dify competes with n8n, LangChain-based stacks and low-code platforms such as Microsoft Power Apps.

Category

AI Agents & Orchestration

Pricing

Freemium, Subscription, Contact for pricing

Target Market

CIOs, CTOs, Heads of AI, Enterprise Developers

Deployment

Open-source, Self-hosted, Cloud-first

Headquarters

Menlo Park, USA

Customers

280+ enterprises; 2,000+ teams on commercial versions (March 2026)

• ✓
  Workflow Studio

  Visual canvas for composing LLM, tool, retrieval and human-input nodes into workflows that move to production without rebuilding the stack.

• ✓
  Knowledge Pipeline (RAG)

  Ingests, chunks and indexes documents and serves retrieval to apps, including hybrid and full-text search on backends such as TiDB Vector.

• ✓
  Dify Agent (open beta)

  Agent builder with a built-in sandbox for code and shell execution and reusable skills, introduced in release 1.16.0 in July 2026.

• ✓
  Model and plugin marketplace

  Connects hundreds of LLMs plus tool and provider plugins, so teams can switch models without rewriting their applications.

• ✓
  MCP support

  Supports the Model Context Protocol 2025-06-18 specification with version negotiation and structured tool output for agent tool use.

• ✓
  Flexible deployment

  Runs as Dify Cloud, a self-hosted or VPC Enterprise edition with SSO, SAML and RBAC, or a free Docker-based Community Edition.

• ✓
  Observability and tracing

  Built-in monitoring plus provider-neutral tracing with Phoenix and LangSmith adapters for debugging workflows running in production.

• •
  Internal knowledge assistants

  Build RAG chatbots over policies, manuals and wikis that employees query through a published web app or an embedded widget.

• •
  Customer support workflows

  Combine retrieval, tools and human-input approval steps into support workflows that draft or resolve tickets with human oversight.

• •
  Business-led AI prototyping

  Let business teams prototype AI apps visually, then hand the same workflow to engineering for production hardening without a rewrite.

• •
  Private self-hosted AI apps

  Run the Enterprise edition inside a VPC so prompts and documents stay within the organisation's own infrastructure and compliance boundary.

• •
  Backend-as-a-service for AI features

  Expose finished workflows as APIs that product teams call from existing applications instead of writing orchestration code.

Dify Cloud publishes list prices: a free Sandbox with 200 message credits, Professional at $590/year (5,000 message credits per month, 3 members) and Team at $1,590/year (10,000 credits per month, 50 members), metered by message credits, apps and knowledge storage. The self-hosted Community Edition is free but limited to a single workspace; SSO, multiple workspaces, advanced security controls and professional support require the custom-priced Enterprise edition.

This page was written from 8 sources, 6 on domains other than dify.ai.

1. 1.dify.ai — dify.aivendor
2. 2.dify.ai — pricingvendor
3. 3.github.com — dify
4. 4.github.com — 1.16.0
5. 5.raw.githubusercontent.com — LICENSE
6. 6.venturebeat.com — dify raises 30 million series pre a to power enterprise grad
7. 7.peerspot.com — dify reviews
8. 8.news.ycombinator.com — item