Dify

Dify is LangGenius's visual agent-workflow and RAG platform. Rechecked 2026-08-17: dify.ai/pricing sells Sandbox, Professional $59 / workspace / month (the annual toggle shows $590 / year), Team $159 / workspace / month ($1,590 / year), and Enterprise custom. github.com/langgenius/dify had 152,668 stars the same day. The LICENSE is a modified Apache 2.0, not a plain Apache-2.0 grant: multi-tenant SaaS and logo/copyright removal need a commercial license. Latest GitHub release we opened was v1.16.1 (2026-07-28). The old 125k-star / 40,000-instance / 10-app Sandbox copy is out.

Compare n8n if you wanted a general automation canvas, Coze if you wanted ByteDance's hosted bot studio, and LangChain if you wanted a code-first framework instead of a canvas.

Key Features

• Visual workflows: Drag LLM, retrieval, branch, HTTP, and code nodes. Treat the live node palette as the catalog; do not freeze "15+ components" as a contract.
• RAG pipeline: Knowledge bases with vector / full-text / hybrid retrieval. Cloud quotas still cap documents and storage per plan.
• Agents and tools: Built-in and custom tools, plus MCP where the current build exposes it. Recheck the tool list in-app.
• Model picker: Marketing still says 200+ models. Use the live provider list, not a 2025 blog count.
• Self-host vs cloud: Community Edition is the $0 self-host path. Cloud is metered by message credits and workspace seats.

Use Cases

• Internal knowledge Q&A that must stay in your VPC.
• Prototype-to-API work where a PM can draw the graph and a backend later wraps it.
• Teams comparing Dify Cloud vs n8n vs LangChain on public dollars, not on anonymous quotes.

Limitation: Sandbox is 200 message credits, 1 member, and 5 apps, not a production quota. The 70% time-saved quotes are vendor marketing; we did not keep them as audited outcomes.

Pricing

dify.ai/pricing on 2026-08-17. Monthly first; the annual toggle is 10x.

If a leftover blog still lists Sandbox at 10 apps or a plain Apache-2.0 license, treat this table and the GitHub LICENSE as the source.

Getting Started

1. Open dify.ai/pricing and start on Sandbox only long enough to see credit burn.
2. Self-host from github.com/langgenius/dify if data must stay on your machines. Read the extra license clauses first.
3. Pay Professional $59 before you assume Team $159 is required.
4. Recheck n8n or LangChain if you already have an automation or code stack.

Frequently Asked Questions

Is Dify still Apache-2.0?

No. The repo LICENSE is a modified Apache 2.0 with multi-tenant and logo conditions.

Is Sandbox still 10 apps?

Not on the page we opened. Live Sandbox is 5 apps.

Same as n8n?

Both are canvases. n8n is general automation. Dify is LLM / RAG / agent workflows with a hosted credit meter.

Alternatives

• n8n: Broader automation, different license and hosting story.
• Coze: Hosted bot studio if you do not want to self-host.
• LangChain: Code-first if the canvas is the bottleneck.

Tips

1. Quote $59 / $159 from the monthly column, or $590 / $1,590 from the annual toggle.
2. Do not paste 125k stars or 40,000 instances into a deck.
3. Recheck message-credit burn before you put Cloud in production.

Conclusion

Dify is a priced cloud workspace plus a modified-Apache self-host tree, not an unmetered 125k-star slogan. Start at dify.ai/pricing, then decide whether n8n or LangChain already covers the workflow you need.