Skip to content

File Uploads, Links, and Web Search in Open WebUI

Open WebUI allows you to enhance your prompts by injecting external data directly into the AI's context window.

1. File Uploads: "Focused Retrieval" vs. "Entire Document"

Simply drag and drop files into a chat or use the + (plus) icon. After uploading a file, click on the file tag/icon directly inside the chat input bar to open its settings dropdown. Here you can choose between two processing modes:

🔍 Focused Retrieval (this is the default)

  • How it works: Open WebUI breaks the document into smaller chunks and extracts only the parts most relevant to your prompt.
  • When to use: Use this for large files (books, long manuals, massive reports) where you only need answers to specific questions. It saves tokens and prevents the AI from getting overwhelmed.

📄 Entire Document

  • How it works: Open WebUI injects the text of the entire document directly into the AI's context window.
  • When to use: Use this for short or medium-sized files (code files, short articles, resumes, legal contracts) where the AI needs to understand the entire context at once, or when you need a full summary or translation.

2. Attaching Specific Web Pages

Simply pasting a URL into the chat works with newer models that support built-in tools — they fetch and read the page automatically. With legacy models, however, this won't work; use the built-in web fetch function instead.

  • How to do it: Type # followed immediately by the URL (e.g., #https://example.com), or click the + (plus) icon and select Attach Webpage. Click on the file tag/icon to set "Focused Retrieval" or "Entire Document".
  • When to use: Use this when you want the AI to read and analyze one specific, known webpage. The URL must be publicly accessible (no paywalls or login screens).
  • How it works: Open WebUI has a built-in search feature, which is configured to use Google.
  • How to use it: Click the Integrations icon next to the chat bar and toggle on Web Search.
  • When to use: Use this when you don't have a specific link, but need the AI to find up-to-date information, news, or live data from the internet using Google's search results before answering.

4. General Pro-Tips

  • Context Limits: If you choose Entire Document, make sure your active LLM has a large enough context window to handle the full text length.
  • Combine them: You can upload a guideline document using Focused Retrieval, attach a website link, and turn on Web Search at the same time for maximum data synthesis.