Skip to content

docs(agents): openwebui — configure the bundled-Ollama model (CON-1566) #202

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
robballantyne wants to merge 1 commit into
base: main
Choose a base branch
from
Open

docs(agents): openwebui — configure the bundled-Ollama model (CON-1566) #202

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
33 changes: 29 additions & 4 deletions external/openwebui/ROOT/etc/vast_agents/openwebui.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,7 +1,32 @@
## Open WebUI (this image)

Runs **Open WebUI** (chat UI — service "Open WebUI") backed by a bundled
**Ollama** server (service "Ollama API", OpenAI-compatible at `/v1`).
**Open WebUI** (a browser chat UI) backed by a **bundled Ollama** server on the same instance —
so this image both *serves* models and gives you a chat over them. base.md applies. Two
services; get their externally callable URLs + token from the manifest (base.md §5, §9):
```
curl -s http://localhost:11111/capabilities/services # both services, with direct_url + state
curl -s http://localhost:11111/capabilities/endpoints # Ollama's OpenAI /v1 base_url + key
```

- Manage models with the `ollama` CLI: `ollama list`, `ollama pull <model>`.
- For the externally callable `base_url` + auth: `curl -s http://localhost:11111/capabilities/endpoints`.
### The model is the thing you configure (start here)

The bundled **Ollama** (supervisor service `ollama`, internal `:21434`) is the model backend;
**Open WebUI** (service `open_webui`, internal `:17500`) is just the front end, already pointed at
it (`OLLAMA_BASE_URL=http://localhost:21434`). Two ways to set the served model:

- **At launch — `OLLAMA_MODEL`** (e.g. `llama3.2`, `qwen2.5:7b`): Ollama pulls it at boot and it
appears in Open WebUI automatically. (`MODEL_NAME` is an accepted alias.)
- **At runtime — the `ollama` CLI:** `ollama pull <model>` / `ollama list` / `ollama rm <model>`;
a newly pulled model shows up in Open WebUI's model selector right away.

Models persist at **`${WORKSPACE}/ollama/models`**, and Open WebUI's own data (accounts, chats) at
**`${WORKSPACE}/data`** — both survive on the workspace.

### Boot order & direct API access

**Open WebUI waits for Ollama to be up *and* the configured model pulled** (`/tmp/.ollama_ready`)
before it starts — so on first boot with a large `OLLAMA_MODEL` the chat UI is intentionally down
while the model downloads. That's the pull, not a fault. Besides the chat UI, Ollama is directly
usable for code: its native `/api` plus an **OpenAI-compatible `/v1`** (chat/completions,
embeddings). Take the `base_url` + key from `/capabilities/endpoints` to point an external client
(or another agent) straight at it.
Loading