Use Dify and OmniDino to build production-ready AI applications.
Connect OmniDino text models, Embedding, Rerank, and tool-calling capabilities to Dify, then quickly build customer service systems, knowledge bases, automated workflows, and business applications with visual Workflows, Chatflows, Knowledge, and Agents.
What is Dify?
Dify is an open-source Agentic Workflow and AI application platform that integrates model providers, workflows, knowledge bases, tools, and application publishing.
Cherry Studio is for chatting. Dify is for building applications.
In Dify, you can embed OmniDino models into a complete business process. After receiving user input, the system can search a knowledge base, evaluate conditions, call HTTP APIs or tools, and then have the model generate the final answer.
Choose Dify Cloud or the self-hosted Community Edition.
Dify Cloud
Best for quick evaluation and lightweight projects, without maintaining databases, queues, plugin runtimes, or upgrades yourself.
Self-hosted Community Edition
Best for teams that need a custom domain, private deployment, data control, and deeper system integration.
How does a typical Dify AI application work?
Prepare Dify and an OmniDino API Key first.
Official Dify Open-Source Project
Includes the Web app, API, Worker, Plugin Daemon, database, and self-hosting deployment files.
OmniDino API Key
Create a dedicated API Key for Dify and configure quotas, groups, and model permissions for your use case.
Install the OpenAI Provider in Dify and connect it to OmniDino.
Dify now uses plugin-based model providers. Menu names may vary slightly between Dify and plugin versions.
Open Model Providers
Open Dify Workspace settings, then go to the model service page under Model Providers, Models, or Integrations.
Install the OpenAI Provider Plugin
Install the official OpenAI Model Provider from the Marketplace or provider list. If it is already installed, open its configuration directly.
Select a Custom API Endpoint
Open the configuration option for a custom address, OpenAI-compatible Endpoint, or Base URL.
Enter OmniDino Settings
Use your OmniDino API Key, set the Base URL to https://api.omnidino.com/v1, and enter the exact model ID.
Save and Verify Credentials
Dify will test the provider credentials. After verification succeeds, add the LLM, Embedding, or Rerank models separately.
Recommended values for an OpenAI-compatible provider.
Core Fields
Depending on the Dify plugin version, this field may appear as API Base, Base URL, Endpoint, or API Endpoint.
/chat/completions path. The Dify provider appends the appropriate endpoint based on the model type.
Provider: OpenAI
API Key: YOUR_OMNIDINO_API_KEY
API Base: https://api.omnidino.com/v1
LLM Model:
Enter the exact OmniDino chat model ID
Embedding Model:
Enter the exact OmniDino Embedding model ID
Rerank Model:
Enter the exact OmniDino Rerank model IDDistinguish at least three model types in Dify.
Language Model
Used for chat, summarization, writing, Workflows, Chatflows, and Agents. Final knowledge-base answers are also typically generated by an LLM.
Text Embedding
Converts document chunks and user questions into vectors. Without Embedding, a semantic knowledge base cannot index or retrieve content correctly.
Rerank
Reorders initial retrieval results to prioritize the content most relevant to the user’s question. It is optional, but often valuable.
Build a minimal chat application with an OmniDino model.
Create a Blank Chatflow
In Studio, click Create App, choose Chatflow, and start from a blank canvas.
Add an LLM Node
Choose the OmniDino provider and a configured text model in the model selector.
Bind User Input
Pass the user question from the Start node into the LLM Prompt, then connect it to the Answer node.
Run a Minimal Test
Enter a simple question, confirm that Dify returns a result, and verify the request in the OmniDino console.
Publish Applications
After confirming that all nodes run without errors, publish the app as a Dify Web App or externally callable application API.
Turn company files into searchable answer sources.
Let the model decide when to call external capabilities.
An Agent is not just a longer conversation. It lets the model call tools autonomously.
In an Agent node, the model can iteratively decide which tools to use and in what order. Common tools include HTTP requests, search, code execution, databases, custom APIs, and knowledge bases.
An OmniDino Key and a Dify App Key serve different purposes.
OmniDino API Key
Stored by the Dify Model Provider and used by Dify to call OmniDino LLM, Embedding, Rerank, or other model services.
Dify App API Key
Generated after a Dify application is published and used by external websites, programs, or servers to call the orchestrated Dify app.
These scenarios are ideal for Dify + OmniDino.
Enterprise Knowledge-Base Support
Answer customer questions from product materials, FAQs, and support documents while citing knowledge sources.
PDF Analysis Assistant
Upload contracts, reports, or manuals to extract key points, risks, and structured information automatically.
Email Classification and Replies
Detect intent, assign priority, generate drafts, and connect to internal systems.
Content Generation Pipeline
Move from keywords to outlines, full drafts, review, translation, and multi-platform output.
Multi-Model Review Workflow
Use different models to generate, critique, and verify content for higher quality on complex tasks.
Internal Business Assistant
Connect CRM, ERP, inventory, quotation, or ticketing systems for queries and automation.
When a connection fails, troubleshoot the model, provider, and nodes layer by layer.
https://api.omnidino.com/v1, the key is valid and has available balance, and the current OpenAI Provider plugin supports a custom Endpoint.Turn OmniDino from a model gateway into the engine behind your AI applications.
Start with a minimal Chatflow containing only Start, LLM, and Answer. Once the connection is stable, add a knowledge base, tools, and Agent nodes.