Use the compatible provider configuration
As a workspace owner or administrator, open Integrations > Model Provider. Install OpenAI-API-compatible by langgenius, then select Setup or Add Model. Older Dify versions may place model providers under Settings. The official plugin listing reviewed on October 4, 2026 shows version 0.0.68.
For an LLM/chat model, configure Type as LLM, Completion mode as Chat, Model Name as your enabled API model ID, API Key as your SinoRouter key, and API Base URL as https://api.sinorouter.ai/v1. This is the /v1 API path under the single gateway https://api.sinorouter.ai/, not a second gateway. Do not paste the complete /chat/completions endpoint into this field.
Set context and maximum output limits to the limits confirmed for your model and account. Only enable streaming, vision, or tool-related capabilities after confirming support. Do not reuse these chat settings for embeddings, reranking, or speech: the plugin can construct those paths differently.
Type: LLM
Completion mode: Chat
Model Name: YOUR_ENABLED_MODEL_ID
API Base URL: https://api.sinorouter.ai/v1
API Key: your workspace key (keep private)Test the model inside a small workflow
Create a simple chat application with a short prompt and a low output limit. Confirm the model can answer before adding knowledge retrieval, tools, or long documents. Inspect the request error and final URL if the connection fails.
Use a model ID returned by GET /v1/models or confirmed in your console. A display name such as a provider family name may not be accepted by the API. Keep the key in Dify’s server-side credential store and never paste it into a public workflow or client-side code.
Before sharing the application
Set application limits, review prompt content, and decide which users can invoke the workflow. Measure token usage and response errors during a small internal test. Image inputs, tools, structured output, and reasoning behavior depend on the selected model and Dify version.
This guide describes configuration structure. It is not a claim that every Dify feature or model has been authenticated on the SinoRouter route. Confirm the features required by your application with support before launch.
- Gateway: https://api.sinorouter.ai/
- Use a key created for the intended workspace.
- Verify the final request path and exact model ID.
- Start with a small, non-sensitive test.
Reference and next checks
Use the applicable provider or application documentation alongside your account-specific route information. These guides explain configuration and evaluation; they do not certify live support for every feature.