Use Dify with llmrelay

Setup

1. 1 In your workspace go to Integrations -> Model Provider and install the OpenAI-API-compatible plugin from the Marketplace.
2. 2 Click Add Model on the plugin card.
3. 3 Set Model Type to LLM, Model Name to a relay model ID (e.g. gpt-5.6-terra), and Completion mode to Chat.
4. 4 Set API Key to your relay key and API Base URL to https://api.llmrelay.dev/v1.
5. 5 Adjust context size and max tokens, then Save; Dify validates the credentials before enabling the model.
6. 6 Set it as System Reasoning Model under Default Models if you want apps to use it by default.

Config

# Dify: Integrations -> Model Provider -> OpenAI-API-compatible -> Add Model
# Model Type:      LLM
# Model Name:      gpt-5.6-terra
# API Key:         ***
# API Base URL:    https://api.llmrelay.dev/v1     # chat/LLM models
#                  https://api.llmrelay.dev        # embedding / rerank / STT / TTS types
# Completion mode: Chat

Smoke test

Copy, paste, run. If you get JSON back with a message, you're good.

Save the model (Dify validates credentials on save), then open any app's Prompt IDE, pick the model and run one test message.

Gotchas

• !For non-LLM model types (text embedding, rerank, STT, TTS) the plugin appends the API version internally, so use the host without /v1 to avoid a doubled /v1/v1 path.
• !A custom model depends on its own credentials; deleting its only key removes the model.
• !Only workspace owners and admins can manage providers, and keys grant workspace-wide model access.

Related integrations