OpenLuxBuild Dify apps on OpenLux models
Dify builds chat apps, agents and RAG pipelines from a model provider list. Add OpenLux as an OpenAI-API-compatible provider and every app you build draws from the same catalogue and the same key.
Set it up in four steps
Create an OpenLux key
Create the key in the console. Dify uses it for chat models, agents and knowledge-base embeddings alike.
Add the compatible provider
In Settings → Model Provider install OpenAI-API-compatible, then set the API endpoint to the OpenLux base URL and paste your key.
Declare the models you use
Add model names exactly as the catalogue spells them, and mark whether each is a chat model or an embedding model.
Build apps and knowledge bases
Chat apps, workflows and document Q&A then run on OpenLux, with embedding cost separated from generation cost in the console.
Configuration
Copy these blocks, replace the placeholder key and you are done.
Settings → Model Provider → OpenAI-API-compatible
API Endpoint: https://openlux.lol/v1
Model Name: gpt-6-astra
Model Type: LLM
Context Size: 128000
(the same provider entry also hosts deepseek-v4, qwen3-max and the embedding model bge-m3)Knowledge base embedding model
Model Name: bge-m3
Model Type: Text Embedding
Endpoint: https://openlux.lol/v1Models that fit Dify
GPT-6 Astra
OpenAI · $0.3677 / 1M tokens
DeepSeek V4
DeepSeek · $0.0613 / 1M tokens
Qwen3 Max
Qwen · $0.0784 / 1M tokens
BGE-M3
AIGC · $0.0035 / 1M tokens
The full list is in the model catalogue and the use-case collections.
Dify + OpenLux questions
Does Dify support custom OpenAI-compatible providers?
Yes. The OpenAI-API-compatible provider plugin exists for exactly that purpose: it takes an API endpoint and a key, then exposes whatever models you declare.
Can the knowledge base use OpenLux embeddings too?
Yes. BGE-M3 is in the catalogue as an embedding model and is metered at $0.0035 per million tokens, so retrieval runs on the same key as generation.
Which model suits a Dify agent?
Agents chain many calls, so cost per token matters. DeepSeek V4 and Qwen3 Max are efficient workhorses; switch to GPT-6 Astra when the reasoning depth is the bottleneck.
How do I see what an app costs?
Usage is metered per model in the OpenLux console. Create a separate key per app if you want the spend separated by product.
Other integration guides
n8n
n8n automates pipelines with a visual editor, and its OpenAI node accepts a custom base URL.
LangChain
LangChain's OpenAI integrations accept a base_url, which makes OpenLux a drop-in upgrade: keep ChatOpenAI, swap the model name to change provider, and run every chain, tool and embedding on one key and one bill..
LobeChat
LobeChat ships with an OpenAI provider that accepts a custom API host.
Connect Dify to OpenLux today
One key covers every model in this guide and the rest of the catalogue, metered in USD with free trial credit to start.