tools
Dify - Open-Source LLM App Builder with RAG & Agents
Dify is the open-source platform for building LLM applications with visual workflows, RAG pipelines, and agent orchestration. Self-host or use cloud.

Dify is a visual platform for creating AI applications with workflows, agents, knowledge pipelines, model providers, tools, and integrations. Apps can be published as web experiences, APIs, embeds, or MCP-compatible tools, with logs and usage monitoring.
Teams can use Dify Cloud, self-host Dify Enterprise, or deploy the Community Edition with Docker. Cloud plans include message credits, while model usage beyond included credits may require users to connect their own provider API keys. Enterprise pricing is custom.
Features
- Build visual workflows with prompts, retrieval, decisions, tools, and human input
- Create agents by chatting or configuring skills, tools, and knowledge
- Extract, clean, chunk, index, and test knowledge from files, websites, and drives
- Install model providers, tools, data sources, and MCP integrations
- Publish apps as hosted experiences, API endpoints, embeds, or MCP-compatible tools
- Monitor logs, feedback, annotations, latency, and usage data
- Self-host with Docker under the Apache-2.0-derivative license
Use cases
- Build customer-support chatbots grounded in company documents
- Create internal agents that retrieve data and call business tools
- Automate multi-step research, routing, and approval workflows
- Prototype and deploy AI applications without building infrastructure
- Create searchable knowledge bases from files, websites, and online documents
Pros
Cons
Latest updates
- v1.19.0 - Trace Your Workflows, Layer by Layer (v1.19.0)
Added nested workflow tracing, portable Workflow Tools in App packages, Skill Hub automation, and easier Agent file upload and preview.
- v1.18.0 — Smarter Agent Routing, Portable Apps, Deeper Workflow Visibility (v1.18.0)
Added Agents that route workflows, portable app packages, nested workflow run visibility, opt-in Agent Knowledge Base, OpenAPI, and difyctl.
- v1.17.0 (v1.17.0)
Adds an E2B sandbox backend, build-time home snapshots, and skill management.
- v1.16.0 (v1.16.0)
Launches Dify Agent (Beta), with a UI builder for agents that run in a Linux sandbox.
Capabilities
- Choice of models — “Message credits are provided to help you easily try out different models from OpenAI, Anthropic, Gemini, xAI, Tongyi in Dify.” source
- Self-hosted — “Source-available. Self-host with Docker.” source
- API — “Publish Dify apps as hosted experiences, API endpoints, embeds, or MCP-compatible tools.” source
- Builds agents and workflows — “Build an agent by chatting or configuring it manually. Use it as an app, embed it into a website, or reuse it as an agent node inside a workflow.” source
- Search over your data — “Create repeatable pipelines for files, websites, online documents, and drives.” source
- Traces and evaluates — “Use logs, feedback, annotations, latency, and usage data to understand what happened and improve the app over time.” source
Get it
Pricing
- Starting price
- $590/yr
- Prices checked
- 2026-10-25
Professional
- $590 per workspace / year
- 5,000 message credits / month
- 1 Team Workspace
- 3 Team Members
- 50 Apps
- 500 Knowledge Documents
- 5GB Knowledge Data Storage
Team
- $1590 per workspace / year
- 10,000 message credits / month
- 1 Team Workspace
- 50 Team Members
- 200 Apps
- 1,000 Knowledge Documents
- 20GB Knowledge Data Storage
Sandbox
Free
- 200 message credits
- 1 Team Workspace
- 1 Team Member
- 5 Apps
- 50 Knowledge Documents
- 50MB Knowledge Data Storage
Enterprise
Price on request
- Enterprise-grade Scalable Deployment Solutions
- Commercial License Authorization
- Multiple Workspaces & Enterprise Management
- SSO
- Advanced Security & Controls
- Professional Technical Support
Community
Free
- All Core Features Released Under the Public Repository
- Single Workspace
- Complies with Dify Open Source License