The history and evolution of Dify, from its origins as an LLM app development tool to becoming one of the most popular open-source LLMOps platforms with over 132,000 GitHub stars.
Dify is an open-source platform for developing LLM applications, combining agentic AI workflows, RAG pipelines, agent capabilities, model management, and observability features. Created by LangGenius, it has become one of the most popular tools for enterprise AI development with over 132,000 GitHub stars and 1,175+ contributors.
Dify was created by LangGenius, a company focused on making LLM application development accessible to both developers and non-developers. The project emerged in early 2023 in response to:
- Growing complexity of LLM application development
- Need for visual workflow orchestration
- Demand for enterprise-ready LLMOps
- Desire for open-source AI infrastructure
The founding mission was to provide a platform that:
- Simplifies LLM application development
- Provides visual workflow building for non-developers
- Supports enterprise deployment and scaling
- Maintains open-source values with sustainable business model
The name “Dify” reflects:
- DIY (Do It Yourself) - Empower users to build their own AI apps
- Modify - Customize and extend AI capabilities
- Simplify - Make LLM development accessible
¶ 2023 - Launch and Foundation
Early 2023:
- Initial Release - First public version of Dify
- Core Features - Basic LLM integration, prompt management
- Community Response - Rapid adoption in developer community
Mid 2023:
- Visual Workflow Builder - Drag-and-drop canvas introduced
- RAG Capabilities - Document ingestion and retrieval
- Model Management - Multi-provider support
Late 2023:
- Agent Framework - Function calling and ReAct patterns
- 50+ Built-in Tools - Google Search, DALL·E, WolframAlpha
- 10,000+ Stars - Major GitHub milestone
¶ 2024 - Growth and Expansion
Q1 2024:
- LLMOps Features - Monitoring and observability
- Backend-as-a-Service - Full API support
- 20,000+ Stars - Continued growth
Q2 2024:
- Enterprise Features - Multi-tenant support, access control
- Kubernetes Support - Helm chart deployment
- 30,000+ Stars - Enterprise adoption
Q3 2024:
- Plugin System - Extensible plugin architecture
- Workflow Improvements - Enhanced orchestration
- 50,000+ Stars - Major community milestone
Q4 2024:
- Advanced RAG - Deep document understanding
- Code Execution Sandbox - Secure code execution
- 70,000+ Stars - Continued adoption
Q1 2025:
- Human-in-the-Loop - Workflow pausing for human review
- Action-Based Routing - Approve, Reject, Escalate flows
- 80,000+ Stars
Q2 2025:
- Celery Architecture - Improved worker scalability
- Redis Cluster Support - Sharded Pub/Sub
- 90,000+ Stars
Q3 2025:
- Enhanced Security - Removed dynamic function evaluation
- Plugin Manifest Pre-caching - Performance improvements
- 100,000+ Stars - Major milestone
Q4 2025:
- Localization - Dutch (nl-NL) language support
- AppList API Optimizations - Query performance
- 115,000+ Stars
February 2026 - Version 1.13.0:
- Human-in-the-Loop (HITL) - Native workflow pausing
- Human Input Node - Review & Edit UI
- Action-Based Routing - Custom buttons for workflow paths
- Celery Worker Updates - New
workflow_based_app_execution queue
- Redis Cluster Mode - Sharded Pub/Sub support
- Security Improvements - Removed dynamic function evaluation
- Dutch Language - nl-NL localization
March 2026 - Version 1.13.2:
- Bug Fixes - Critical regression fixes from v1.13.1
- Prompt Message Fixes - LLM-plugin invocation issues resolved
- Knowledge Retrieval - Node execution failures fixed
- Weaviate Client - Cleanup issues resolved
- Release Date: March 18, 2026
| Date |
Milestone |
| Early 2023 |
Initial Dify release |
| Mid 2023 |
Visual workflow builder launched |
| Late 2023 |
Agent framework with 50+ tools |
| Late 2023 |
10,000 GitHub stars |
| Q1 2024 |
LLMOps features released |
| Q2 2024 |
Enterprise features added |
| Q3 2024 |
Plugin system launched |
| Q3 2024 |
50,000 GitHub stars |
| Q4 2024 |
Advanced RAG capabilities |
| Q4 2024 |
70,000 GitHub stars |
| Q1 2025 |
Human-in-the-Loop feature |
| Q3 2025 |
100,000 GitHub stars |
| Q4 2025 |
Dutch language support |
| Feb 2026 |
Version 1.13.0 with HITL |
| Mar 2026 |
Version 1.13.2 (March 18, 2026) |
| 2026 |
134k+ stars, 1,175+ contributors |
Early Versions:
- Simple Flask backend
- Basic React frontend
- Single database (PostgreSQL)
Current Versions:
- Microservices architecture
- Next.js frontend
- PostgreSQL + Redis
- Celery workers for async tasks
- Code execution sandbox
- Plugin daemon
Evolution:
- Simple Prompts - Basic prompt management
- Visual Builder - Drag-and-drop workflow canvas
- Agent Framework - Function calling and ReAct
- Human-in-the-Loop - Workflow pausing for human review
- Action-Based Routing - Custom workflow paths
Evolution:
- Basic Retrieval - Simple vector search
- Hybrid Search - Vector + keyword search
- Reranking - BGE-Reranker integration
- Deep Document Understanding - PDF, PPT, complex formats
- Multi-Document - Cross-document retrieval
Evolution:
- OpenAI Only - Initial GPT integration
- Multi-Provider - Anthropic, Google, Mistral
- Local Models - Ollama, llama.cpp
- Hundreds of LLMs - Comprehensive model marketplace
- Apache 2.0 based
- Commercial use allowed
- Some restrictions on SaaS
Dify Open Source License (Apache 2.0 with additional conditions):
Allowed WITHOUT commercial license:
- ✅ Commercial use as backend service
- ✅ Enterprise application development
- ✅ Internal business use
- ✅ Modified versions for internal use
Requires commercial license:
- ❌ Multi-tenant SaaS offerings
- ❌ Remove LOGO/copyright from console or apps
Contributor Agreement:
- Contributors agree LangGenius can adjust license
- Contributions may be used for commercial purposes
¶ Community and Ecosystem
| Metric |
Value |
| Stars |
134,000+ |
| Forks |
20,900+ |
| Watchers |
774 |
| Contributors |
1,175+ |
| Releases |
160 |
| Issues |
Active triage |
| Pull Requests |
Active review |
| Metric |
Value |
| GitHub Stars |
134k+ |
| Forks |
20.9k+ |
| Contributors |
1,175+ |
| Releases |
160 |
Supported Integrations:
- LLM Providers: OpenAI, Anthropic, Google, Mistral, Llama3
- Local Models: Ollama, llama.cpp
- Vector Databases: Weaviate, Milvus, Chroma, Qdrant
- Tools: 50+ built-in tools (Google Search, DALL·E, WolframAlpha)
- Cloud: AWS, Azure, GCP, Kubernetes
Dify has significantly influenced the LLM application development space:
- Accessibility - Made LLM app development accessible to non-developers
- Visual Orchestration - Popularized visual AI workflow building
- Enterprise Focus - Focused on production deployment needs
- Open Source - Apache-2.0 based license encourages adoption
- LLMOps - Advanced monitoring and observability features
- 132k+ GitHub stars (one of the most starred AI projects)
- Widely adopted in enterprise deployments
- Featured in AI and enterprise software communities
- Recommended for LLM application development
- Used for customer support bots, knowledge management, workflow automation
- Latest Version: v1.13.2 (March 18, 2026)
- Release Cadence: Regular monthly releases
- Active Development: Continuous feature additions
- Bug Tracking: Public GitHub issues
- Community: Active Discord and GitHub discussions
- Visual workflow orchestration
- Human-in-the-Loop (HITL)
- Agent framework with 50+ tools
- Advanced RAG with deep document understanding
- LLMOps monitoring and observability
- Backend-as-a-Service with full API
- Code execution sandbox
- Plugin system
- Multi-language support (including Dutch)
Open-Source:
- Dify Open Source License (Apache 2.0+)
- Free for most use cases
- Community-driven development
Commercial License:
- Required for multi-tenant SaaS
- Required for white-label deployments
- Contact LangGenius for licensing
Based on development patterns and public communications:
- Enhanced Human-in-the-Loop features
- Improved workflow debugging
- Better plugin ecosystem
- More language localizations
- Enhanced mobile experience
- Maintain open-source license
- Expand enterprise features
- Enhanced collaboration tools
- Better multi-tenancy support
- Improved analytics and monitoring
- More LLM provider integrations
- Focus: LLM application development platform
- Products: Dify (open-source), Dify Cloud (managed service)
- Location: Global operations
- Mission: Simplify LLM application development
- Website: https://dify.ai
- GitHub: https://github.com/langgenius/dify
- Documentation: https://docs.dify.ai
- Cloud: https://cloud.dify.ai
- Discord: https://discord.gg/dify
- Twitter: @dify_ai
Dify represents an enterprise-focused approach to LLM application development. Its evolution from simple prompt management to a comprehensive platform with visual workflow orchestration, Human-in-the-Loop, and advanced LLMOps reflects the broader trajectory of enterprise AI adoption.
Key principles that guide Dify:
- Enterprise-Ready - Focused on production deployment
- Visual Orchestration - Easy workflow building for all users
- Open Source - Apache-2.0 based license (with conditions)
- LLMOps Excellence - Best-in-class monitoring and observability
- Flexibility - Support for hundreds of LLM providers
Setting up LLM applications can be complex. We offer consulting services for:
- Production-ready deployments
- Performance tuning and optimization
- High-availability configurations
- Migration from legacy systems
Contact us at office@linux-server-admin.com or visit our contact page.