Successfully Running Open WebUI + Ollama on Windows 11 (Docker)

0Orientation & objectives

This module gets one specific machine — Windows 11, Intel Core Ultra 265K, ASUS Z890-P WiFi, Intel Arc B570 Challenger GPU, 32GB DDR5 — from zero to a working Open WebUI + Ollama stack in Docker, and flags the two places where “successfully running” on this hardware isn’t fully answered by current documentation: whether GPU acceleration should happen inside Docker or outside it, and how to tell which one is actually working.

Learning objectives. On completion a participant will be able to describe what Open WebUI and Ollama each do and how they connect, explain why Docker Desktop on Windows 11 depends on WSL2 for GPU access, stand up both services in Docker with persistent data and correct networking, describe the two documented (but individually unconfirmed) paths to GPU acceleration on Intel Arc and how to test which one is running, and apply the update procedure and Windows-specific fixes without losing data.

Element	Detail
Format	Self-paced; approximately 25–35 minutes
Formative questions	17 multiple-choice, distributed (feedback, not scored)
Final assessment	12 multiple-choice questions; pass mark 75% (9/12)
Sources	This repository’s wiki (official documentation, 2026-07-07/08) + external sources for Intel Arc GPU support (fetched 2026-07-13, cited by URL)

Related but distinct. This module overlaps by design with openwebui-stack-docker-course (general, cross-platform). This one is scoped to Windows 11 and this specific hardware, where a general course can’t go deep enough.

1The stack on Windows 11

Learning objective. State what Open WebUI and Ollama each do, describe how they connect, and identify each project’s primary install path on Windows specifically.

Open WebUI is an extensible, self-hosted AI platform designed to run entirely offline, working with both Ollama and OpenAI-compatible APIs rather than locking users into one model provider. (source: openwebui-intro)

Ollama is local large-language-model infrastructure built on llama.cpp. Its own scope stays narrow — model serving — while a large third-party ecosystem, including Open WebUI, builds interfaces and tooling around it. (source: ollama-readme)

The two connect asymmetrically: Open WebUI’s OLLAMA_BASE_URL (default http://localhost:11434, different inside Docker networking) points at a running Ollama server, and Open WebUI’s backend proxies straight through to Ollama’s native API under /ollama/<api>. Ollama, for its part, is not aware of Open WebUI at all — it lists Open WebUI only as one more community front end in its own documentation. (source: concept-integration-openwebui-ollama)

Install paths differ by platform

	Open WebUI	Ollama
Primary path	Docker (all platforms)	Native installer on Windows (curl script on macOS/Linux)
Alternative	pip, uv, native desktop app	Docker image
Default port	3000 (Docker)	11434

Open WebUI’s primary, most-supported path is Docker (ghcr.io/open-webui/open-webui:main) on every platform. Ollama’s primary path is platform-dependent: a curl-piped script on macOS/Linux, but a native installer (OllamaSetup.exe) on Windows — Docker (ollama/ollama) is documented as the alternative, not the default, on Windows. (source: concept-deployment-docker, ollama-windows)

This course runs both services in Docker throughout, as scoped. Section 4 revisits why the native install remains the more reliably documented choice specifically for GPU acceleration on this hardware.

Check your understanding

1.1 What does Open WebUI’s OLLAMA_BASE_URL variable do?

1.2 On Windows specifically, what is Ollama’s primary, most-documented install path?

1.3 Which of the two projects is aware of the other’s specific configuration?