Run Dify at 55% Lower Cost
Move self-managed Dify workflows from local Docker to persistent CPU infrastructure with public endpoints, background workers, and storage. Connect the model provider you choose—or a separate self-hosted inference endpoint—while keeping the Dify application stack on CPU.
Built for always-on Dify hosting
Dify’s visual canvas is only one part of the stack. Give its apps, APIs, workers, schedules, databases, and vector state a long-lived home beyond your local Docker session.
Publish Dify web apps and REST APIs through a reachable public endpoint. Configure a public trigger URL so external systems can invoke your webhook and integration workflows.
Workflows that keep working
Run scheduled and event-driven Workflow apps without depending on a laptop that sleeps, disconnects, or reboots.
Start from $9 per month
Use flat daily infrastructure pricing and zero egress fees for the Fluence compute portion of your stack.
One architecture for Dify workflow and agents
Run Dify’s application, orchestration, and data services on Fluence CPU Cloud. Keep model inference outside the base VM by default, or add a separate GPU endpoint when local models require one.
Run on Fluence CPU Cloud
Dify web console and published applications
REST API and WebSocket service Background workers and workflow scheduler PostgreSQL and Redis Vector database and file storage Plugin daemon, reverse proxy, and sandboxes
Connect externally
Hosted LLM, embedding, and reranking providers
Custom or self-hosted model endpoints
Optional Fluence GPU inference endpoint
Email, CRM, search, and other business tools
DNS, TLS, backup, and monitoring services
Choose a region, server type, storage, operating system, and networking settings for your Dify backend.
Install Docker and Docker Compose, clone a tagged Dify release, and copy the example environment file:
cd dify/docker
cp .env.example .env
Replace the default secrets, configure public URLs, and start the stack:
docker compose up -d
Add a model-provider key or custom model endpoint, publish your applications and triggers, and verify that every required container is healthy.
Before production, configure TLS, monitoring, backups, restore tests, firewall rules, certificate renewal, and a controlled upgrade process.
Use lower-cost CPU infrastructure for the Dify runtime, then add inference, storage, and external services only where the workload requires them.
No egress fees
Transparent daily pricing
Public IPv4 and persistent storage available
Note: Calculator estimates reflect only the Virtual Server layer used for LlamaIndex hosting.
What you can build with Dify on Fluence
Knowledge-grounded customer support
Build a support assistant that retrieves answers from product documentation, policies, and internal knowledge using embeddings and a vector database.
Intent-based email routing
Classify incoming support messages and send each one to the correct queue, team, or downstream workflow while keeping the routing logic visible and auditable.
Scheduled reports and notifications
Run recurring workflows that gather data, generate summaries, and send daily reports, briefings, alerts, or notifications on a timetable.
Agentic RAG and research
Give an agent access to retrieval and tools so it can plan, search, retry, use fallbacks, and synthesize results across multi-step questions.
A global marketplace of compute
Compare available providers and locations, provision the CPU instance for your Dify runtime, attach storage and public IPs, and add GPU resources only when self-hosted inference requires them.
FAQ
Can I self-host Dify on Fluence?
How much CPU and RAM does Dify need?
Does Dify need a GPU?
Which model providers can I connect to Dify?
Can Dify expose APIs, webhooks, and scheduled workflows?
Is Docker supported, and what should I configure before production?
What does the $10.78/month Fluence price include?
Can I use Dify Community Edition commercially?
