No-code tools
Anything that can talk to an OpenAI-compatible endpoint can talk to Djelia. The settings are always the same three:
| Setting | Value |
|---|---|
| Base URL / API base | https://api.djelia.cloud/openai/v1 |
| API key | your Djelia key |
| Model | banjugu-1, sunjata-1 or jifili-1 |
Open WebUI
Settings → Connections → OpenAI API. Set the base URL and key above. Djelia's models
appear in the model picker, since /models is implemented.
Translation is a chat model, so it works in the chat interface, but it needs a language pair. Open WebUI has no field for arbitrary body parameters, so put Djelia behind a LiteLLM proxy with the pair configured there, and point Open WebUI at the proxy.
n8n
Use the OpenAI node with a custom credential: set the base URL to
https://api.djelia.cloud/openai/v1 and paste your key. For the language pair, use the
HTTP Request node instead when the OpenAI node does not expose custom body fields.
Dify
Model Provider → OpenAI-API-compatible. Add the base URL and key, then declare each Djelia model you want with its capability (LLM for translation, Speech2Text for ASR, TTS for speech).
Anything else
If a tool asks for an "OpenAI-compatible base URL", it will work. The one thing to check
is whether it lets you send extra body parameters, which is what the language pair for
translation needs. If it does not, a LiteLLM proxy in front of
Djelia solves it, because the pair can live in config.yaml instead.
Transcription and speech need no extensions, so they work anywhere without a proxy.