Open WebUI with unoblox (India)

Open WebUI is the self-hosted chat interface many teams put in front of their own model backend instead of a public chatbot. Point its OpenAI connection at unoblox and your organisation gets a private, chat-style front end backed by GPT, Claude, DeepSeek, Qwen, and more, billed in rupees on one monthly GST invoice.

Why teams put Open WebUI in front of unoblox

Open WebUI already speaks the OpenAI API format for chat, so it does not need a special plugin to work with unoblox — it needs a base URL and a key, exactly as it would for any OpenAI-compatible host. The result is a private workspace: logins, chat history, and model access controls live in your own Open WebUI instance, while every message is billed centrally through your unoblox account instead of many individual staff subscriptions.

Connecting Open WebUI

In Open WebUI, add a new OpenAI API connection under admin settings:

• API base URL: https://api.unoblox.ai/v1
• API key: ub-gw-xxxxxxxxxxxxxxxx

If you run Open WebUI via Docker, the same two values go into the container's environment:

docker run -d -p 3000:8080 \
  -e OPENAI_API_BASE_URL=https://api.unoblox.ai/v1 \
  -e OPENAI_API_KEY=ub-gw-xxxxxxxxxxxxxxxx \
  -v open-webui:/app/backend/data \
  --name open-webui ghcr.io/open-webui/open-webui:main

Once it restarts, every model in your unoblox catalog shows up in Open WebUI's model picker automatically.

Choosing which models to expose to your team

Open WebUI lets you show or hide specific models per user group, which matters once cost is shared across a whole company:

Use in the workspace	Model	₹ per 1M tokens (input / output)
Default for everyone, near-zero cost	Qwen3 1.7B	₹0
Everyday drafting	DeepSeek V4 Flash	₹9.07 / ₹18.14
General-purpose chat	GPT-5	₹126 / ₹1008
Reserved for hard problems	Claude Opus	₹504 / ₹2520

Restrict the expensive tier to a smaller group and let everyone else default to the free or low-cost models — Open WebUI's per-group model visibility handles this without any code.

What stays local, what does not

Open WebUI itself, meaning chat history, user accounts, and uploaded files, runs entirely on infrastructure you control, since it is self-hosted. The model call itself still goes out to whichever provider you selected: GPT, Claude, and similar external models process it on their own infrastructure outside India, with unoblox adding rupee billing, a GST invoice, and one endpoint rather than India data residency. Only the small unoblox-hosted models in the catalog actually run on India infrastructure.

Frequently asked questions

Do we need one unoblox key per employee? No — most teams use one organisation key behind Open WebUI and rely on Open WebUI's own user accounts and roles for per-person access control.

Can we limit which models junior staff can pick? Yes, Open WebUI supports per-group model access; pair that with the pricing table above to keep the expensive models reserved for the people who need them.

Does this support file uploads and retrieval inside Open WebUI? Open WebUI's document and retrieval features work independently of which chat backend you point it at, so they continue to function; the chat completion call itself still goes through the unoblox endpoint.

Will our chat history be billed as tokens every time? Yes — like any OpenAI-compatible backend, the conversation history sent with each turn counts as input tokens, so longer chat threads cost proportionally more per message.

Can we self-host Open WebUI fully offline? The interface and its data can run offline, but the chat completion itself is an API call to unoblox over the internet, so model calls need connectivity even if the UI does not.

Get started in rupees → https://unoblox.ai/sign-in