First Boot: Ollama + Open WebUI on a Box You Own
First boot of Ollama + Open WebUI on a box you own: install Ollama (credit: the Ollama project), pull one licensed open-weight model, put Open WebUI (credit: that project) on localhost with a password. D-Central does this for you on a Strix Halo 128 GB at $7,349 CAD or a DGX Spark at $9,449 CAD. A discrete tower is a quoted GPU rig. We did not write those programs.
Plain HowTo without buying hardware: Ollama in 10 minutes, Open WebUI.
On any Linux box you already have
- Install Ollama from the project’s current Linux instructions. Do not copy a random curl | sh from a blog older than the release.
ollama pulla model whose licence you accept. Open-weight is not automatically Apache 2.0. Dated shortlist.- Run Open WebUI in Docker or the project’s documented binary, pointed at Ollama’s local API.
- Create an admin user before you leave it on a LAN.
- Do not port-forward it to the world on day one.
On the machines we sell
- Spark ($9,449): DGX OS, Arm64, CUDA. See Spark setup and Ollama on Spark.
- Strix Halo ($7,349): x86, Windows or Linux, ROCm/Vulkan. Same UI, different GPU stack. CUDA vs ROCm.
- GPU rig: we install the same open stack after the card is burned in.
No invented tokens/s. No hosted SLA. No claim that a local UI is Law 25 compliance.
Related products, repair, and setup paths
- self-hosted AI for Bitcoiners hub
- plebs guide to self-hosted AI
- install Ollama in 10 minutes
- LM Studio vs Ollama vs llama.cpp
- connect local AI to Home Assistant and Obsidian
- self-hosted AI troubleshooting
- repurpose mining hardware into an AI hashcenter
- local AI model leaderboards
Last reviewed August 24, 2026.
