docs: explain Create a model and pulling GGUF models from Hugging Face on the Ollama page

[silentoplayz]

Summary

The Ollama page covered pulling, updating and deleting models from the Manage dialog, but not its Create a model section, and it did not say that Ollama can download GGUF models from Hugging Face. This adds both:

• Downloading GGUF Models from Hugging Face: entering hf.co/{username}/{repository}:{quantization} in Pull a model from Ollama.com or the model selector, with an example and a link to Hugging Face's Ollama guide.
• Creating a Model: the name and JSON fields, which JSON fields Ollama accepts (with a link to Ollama's create API), what success and the common errors look like, and a note that a system prompt set this way only applies when the chat sends none of its own.

The experimental Upload a GGUF model section is left out on purpose: it currently creates no model in either mode (open-webui/open-webui#31861). The Hugging Face section covers the same need.

Related issue or discussion

None. The two sections were the remaining undocumented parts of the Ollama Manage dialog after #1448.

Checklist

• I have reviewed the relevant documentation and matched the existing style.
• This PR meets Open WebUI's contribution standards: it is accurate, relevant to users, narrowly scoped, maintainable, and not promotional content, advertising, lead generation, SEO placement, or a request to list a product, service, provider, integration, gateway, tool, or company primarily for visibility.
• I understand that PRs that do not meet these standards may be closed without review and will not be merged. Repeated, low-quality, off-topic, promotional, or intentionally misleading submissions may result in the contributor being blocked from future participation in Open WebUI repositories.

Notes for reviewers

Checked against dev at 7e6c8d7b6 (ManageOllama.svelte and the Ollama router) and tested on Ollama 0.35.0:

• Create a model with {"from": "tiny-gemma", "system": "You always talk like a pirate."} showed success and the model was listed as pirate-gemma:latest. The example JSON on the page (from, system, parameters) was checked against the same server, and ollama show --modelfile showed its system prompt and parameters.
• With from set to a model the server did not have, Ollama answered pull model manifest: file does not exist.
• The created model followed its system prompt in a request without a system message, and ignored it in a request that sent one, which is what the note describes.
• hf.co/ggml-org/gemma-3-270m-it-GGUF:Q8_0 pulled and was listed under that name. Both the Manage pull field and the model selector strip a leading ollama run or ollama pull and pass the rest to Ollama unchanged.