[GH-ISSUE #11323] issue: Web Search Failing in latest 2 updates #16185

Closed
opened 2026-04-19 22:12:06 -05:00 by GiteaMirror · 8 comments
Owner

Originally created by @LazyDataScientistGit on GitHub (Mar 6, 2025).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/11323

Check Existing Issues

  • I have searched the existing issues and discussions.

Installation Method

Git Clone

Open WebUI Version

v0.5.20 thru v0.5.18

Ollama Version (if applicable)

N/A

Operating System

Linux Mint 22

Browser (if applicable)

N/A

Confirmation

  • I have read and followed all instructions in README.md.
  • I am using the latest version of both Open WebUI and Ollama.
  • I have checked the browser console logs.
  • I have checked the Docker container logs.
  • I have listed steps to reproduce the bug in detail.

Expected Behavior

Search results to come back and not fail.

Actual Behavior

All search queries fail and model instructs me that it does not have real-time data.

Steps to Reproduce

Open New Chat,
select any model (I am using openrouter for models)
select "web-browsing"
Ask a query.

Logs & Screenshots

data:
action: "web_search"
description: "Error searching "{{searchQuery}}""
done: true
error: true
query: "2025 top 5 Antidepressants"

Additional Information

I have tried using Qwen2.5-72B-Instruct, Mistral Small 3, and llama3.3-70b-instruct and all have the same issue.

Originally created by @LazyDataScientistGit on GitHub (Mar 6, 2025). Original GitHub issue: https://github.com/open-webui/open-webui/issues/11323 ### Check Existing Issues - [x] I have searched the existing issues and discussions. ### Installation Method Git Clone ### Open WebUI Version v0.5.20 thru v0.5.18 ### Ollama Version (if applicable) N/A ### Operating System Linux Mint 22 ### Browser (if applicable) N/A ### Confirmation - [x] I have read and followed all instructions in `README.md`. - [x] I am using the latest version of **both** Open WebUI and Ollama. - [x] I have checked the browser console logs. - [x] I have checked the Docker container logs. - [x] I have listed steps to reproduce the bug in detail. ### Expected Behavior Search results to come back and not fail. ### Actual Behavior All search queries fail and model instructs me that it does not have real-time data. ### Steps to Reproduce Open New Chat, select any model (I am using openrouter for models) select "web-browsing" Ask a query. ### Logs & Screenshots data: action: "web_search" description: "Error searching \"{{searchQuery}}\"" done: true error: true query: "2025 top 5 Antidepressants" ### Additional Information I have tried using Qwen2.5-72B-Instruct, Mistral Small 3, and llama3.3-70b-instruct and all have the same issue.
GiteaMirror added the bug label 2026-04-19 22:12:06 -05:00
Author
Owner

@rgaricano commented on GitHub (Mar 7, 2025):

what are your settings for websearches?

<!-- gh-comment-id:2705217534 --> @rgaricano commented on GitHub (Mar 7, 2025): what are your settings for websearches?
Author
Owner

@LazyDataScientistGit commented on GitHub (Mar 7, 2025):

@rgaricano just the default duckduckgo setting and i get the same issue with the default google_pse settings.

<!-- gh-comment-id:2705233390 --> @LazyDataScientistGit commented on GitHub (Mar 7, 2025): @rgaricano just the default `duckduckgo` setting and i get the same issue with the default `google_pse` settings.
Author
Owner

@rgaricano commented on GitHub (Mar 7, 2025):

Image

<!-- gh-comment-id:2705256958 --> @rgaricano commented on GitHub (Mar 7, 2025): ![Image](https://github.com/user-attachments/assets/fcbf8750-985a-4945-833b-4816c1b0cae8)
Author
Owner

@raydoomed commented on GitHub (Mar 7, 2025):

I haven't been able to do it recently. I should have bypassed the embedding problem. Anyway, I can't use online search now.

<!-- gh-comment-id:2705277971 --> @raydoomed commented on GitHub (Mar 7, 2025): I haven't been able to do it recently. I should have bypassed the embedding problem. Anyway, I can't use online search now.
Author
Owner

@rgaricano commented on GitHub (Mar 7, 2025):

user group permissions, interface web search query generation ?

Image

Image

Image

<!-- gh-comment-id:2705342944 --> @rgaricano commented on GitHub (Mar 7, 2025): user group permissions, interface web search query generation ? ![Image](https://github.com/user-attachments/assets/7cf3e7ae-d83c-4efe-b612-d72d02af4390) ![Image](https://github.com/user-attachments/assets/590621e1-da2d-457e-a6bc-6603ad025181) ![Image](https://github.com/user-attachments/assets/40febfdb-c085-488b-9aea-718761885988)
Author
Owner

@LazyDataScientistGit commented on GitHub (Mar 7, 2025):

@rgaricano I have all the settings matching your settings and web searches still fail.

<!-- gh-comment-id:2705361533 --> @LazyDataScientistGit commented on GitHub (Mar 7, 2025): @rgaricano I have all the settings matching your settings and web searches still fail.
Author
Owner

@1over137 commented on GitHub (Mar 7, 2025):

I don't get errors per se, but if I bypass the embedding, the model says it has no information. In the logs it shows that the search (searxng) is working. If I use the embedding, it takes forever probably because it is continuously spamming ollama with calls to /api/embed one character at a time.

<!-- gh-comment-id:2705384562 --> @1over137 commented on GitHub (Mar 7, 2025): I don't get errors per se, but if I bypass the embedding, the model says it has no information. In the logs it shows that the search (searxng) is working. If I use the embedding, it takes forever probably because it is continuously spamming ollama with calls to `/api/embed` one character at a time.
Author
Owner

@raydoomed commented on GitHub (Mar 7, 2025):

I don't get errors per se, but if I bypass the embedding, the model says it has no information. In the logs it shows that the search (searxng) is working. If I use the embedding, it takes forever probably because it is continuously spamming ollama with calls to /api/embed one character at a time.

Yes, and you can't use LM studio either. Use ollama to open bypass embedding.

<!-- gh-comment-id:2705507093 --> @raydoomed commented on GitHub (Mar 7, 2025): > I don't get errors per se, but if I bypass the embedding, the model says it has no information. In the logs it shows that the search (searxng) is working. If I use the embedding, it takes forever probably because it is continuously spamming ollama with calls to `/api/embed` one character at a time. Yes, and you can't use LM studio either. Use ollama to open bypass embedding.
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/open-webui#16185