I have searched the existing issues and discussions.
I am using the latest version of Open WebUI.
Installation Method
Docker
Open WebUI Version
v0.6.22
Ollama Version (if applicable)
0.11.5
Operating System
Ubuntu 24.04
Browser (if applicable)
Firefox 141.0.3
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 included the browser console logs.
I have included the Docker container logs.
I have provided every relevant configuration, setting, and environment variable used in my setup.
I have clearly listed every relevant configuration, custom setting, environment variable, and command-line option that influences my setup (such as Docker Compose overrides, .env values, browser settings, authentication configurations, etc).
I have documented step-by-step reproduction instructions that are precise, sequential, and leave nothing to interpretation. My steps:
Start with the initial platform/version/OS and dependencies used,
Specify exact install/launch/configure commands,
List URLs visited, user input (incl. example values/emails/passwords if needed),
Describe all options and toggles enabled or changed,
Include any files or environmental changes,
Identify the expected and actual result at each stage,
Ensure any reasonably skilled user can follow and hit the same issue.
Expected Behavior
by selecting search button, I shall see that model can search through internet and has access to the real-time data on web similar to Perplexity or DeepSeek Search, ...
by clicking on the speaker icon below each reply, I shall hear the text turned into speech
Actual Behavior
almost, everytime, I click on the search button, it replies after thinking for a while either "I do not have access to data after September 21" or saying "an error occured"
I have tried with DuckDuckGo and TavilyAI and both failed.
I have selected local algorithm for both text-to-speech and vice versa. Speech-to-text works well! but by clicking on the icon for text-to-speech, nothing happens. I just gets stuck in showing three dots.
I have a PhD in AI and can help to provide algorithms and their codes for something as similar as possible to the one from Perplexity and also Text-to-Speech, but I do not have background in website and UI development.
Searching is also very slow.
Steps to Reproduce
start with Ubuntu 24.04
User Docker version 28.3.3
Install Open-WebUI using the following docker-compose
Originally created by @yekruz on GitHub (Aug 22, 2025).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/16804
### Check Existing Issues
- [x] I have searched the existing issues and discussions.
- [x] I am using the latest version of Open WebUI.
### Installation Method
Docker
### Open WebUI Version
v0.6.22
### Ollama Version (if applicable)
0.11.5
### Operating System
Ubuntu 24.04
### Browser (if applicable)
Firefox 141.0.3
### 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 included the browser console logs.
- [x] I have included the Docker container logs.
- [x] I have **provided every relevant configuration, setting, and environment variable used in my setup.**
- [x] I have clearly **listed every relevant configuration, custom setting, environment variable, and command-line option that influences my setup** (such as Docker Compose overrides, .env values, browser settings, authentication configurations, etc).
- [x] I have documented **step-by-step reproduction instructions that are precise, sequential, and leave nothing to interpretation**. My steps:
- Start with the initial platform/version/OS and dependencies used,
- Specify exact install/launch/configure commands,
- List URLs visited, user input (incl. example values/emails/passwords if needed),
- Describe all options and toggles enabled or changed,
- Include any files or environmental changes,
- Identify the expected and actual result at each stage,
- Ensure any reasonably skilled user can follow and hit the same issue.
### Expected Behavior
1) by selecting search button, I shall see that model can search through internet and has access to the real-time data on web similar to Perplexity or DeepSeek Search, ...
2) by clicking on the speaker icon below each reply, I shall hear the text turned into speech
### Actual Behavior
1) almost, everytime, I click on the search button, it replies after thinking for a while either "I do not have access to data after September 21" or saying "an error occured"
I have tried with DuckDuckGo and TavilyAI and both failed.
<img width="1409" height="582" alt="Image" src="https://github.com/user-attachments/assets/90e4c231-96f9-4f14-a220-d6d06c1826e8" />
<img width="1409" height="582" alt="Image" src="https://github.com/user-attachments/assets/4255b8ca-09e4-4369-9552-05e237f88d1b" />
<img width="1414" height="1081" alt="Image" src="https://github.com/user-attachments/assets/58d044a9-5354-4f0c-b0e0-90684720ebb3" />
<img width="1154" height="155" alt="Image" src="https://github.com/user-attachments/assets/791c7363-4ae6-483a-bbef-2880caee1ed3" />
<img width="1414" height="1081" alt="Image" src="https://github.com/user-attachments/assets/49d84773-d696-4936-ab9b-0fcefea6ddf9" />
2) I have selected local algorithm for both text-to-speech and vice versa. Speech-to-text works well! but by clicking on the icon for text-to-speech, nothing happens. I just gets stuck in showing three dots.
<img width="236" height="75" alt="Image" src="https://github.com/user-attachments/assets/eb5f2064-9f7c-4fe3-8913-4280f856d140" />
I have a PhD in AI and can help to provide algorithms and their codes for something as similar as possible to the one from Perplexity and also Text-to-Speech, but I do not have background in website and UI development.
Searching is also very slow.
### Steps to Reproduce
1) start with Ubuntu 24.04
2) User Docker version 28.3.3
3) Install Open-WebUI using the following ` docker-compose`
```bash
services:
ollama:
volumes:
- ollama:/root/.ollama
container_name: ollama
pull_policy: always
tty: true
restart: unless-stopped
image: ollama/ollama:${OLLAMA_DOCKER_TAG-latest}
open-webui:
build:
context: .
args:
OLLAMA_BASE_URL: '/ollama'
dockerfile: Dockerfile
image: ghcr.io/open-webui/open-webui:${WEBUI_DOCKER_TAG-main}
container_name: open-webui
volumes:
- open-webui:/app/backend/data
depends_on:
- ollama
ports:
- ${OPEN_WEBUI_PORT-3000}:8080
environment:
- 'OLLAMA_BASE_URL=http://ollama:11434'
- 'WEBUI_SECRET_KEY='
extra_hosts:
- host.docker.internal:host-gateway
restart: unless-stopped
volumes:
ollama: {}
open-webui: {}
```
4) add d`eepseek-r1:32` or `gpt-oss:20`
5-1) configure the search setting and activate search and do a query
5-2) click on the speaker icon below each reply
### Logs & Screenshots
[logs_docker_ollama-webui.txt](https://github.com/user-attachments/files/21933341/logs_docker_ollama-webui.txt)
### Additional Information
_No response_
GiteaMirror
added the bug label 2026-05-18 02:52:41 -05:00
To fix the datetime issue of the model, try adding to system prompt:
Current time: {{CURRENT_TIME}}.
Current date: {{CURRENT_DATE}}.
for search engines
duckduckgo search engine have issues with concurrent request in some of the search engines that it use, it's improved in last version (now appear as DDGS) to reflect the max concurrent request value,
for tabily, as I know, it use max concurrent request, maybe you can do tests with low values to check it funcionality.
<!-- gh-comment-id:3213643186 -->
@rgaricano commented on GitHub (Aug 22, 2025):
To fix the datetime issue of the model, try adding to system prompt:
```
Current time: {{CURRENT_TIME}}.
Current date: {{CURRENT_DATE}}.
```
for search engines
duckduckgo search engine have issues with concurrent request in some of the search engines that it use, it's improved in last version (now appear as DDGS) to reflect the max concurrent request value,
for tabily, as I know, it use max concurrent request, maybe you can do tests with low values to check it funcionality.
To be honest, I've hit a wall with your suggestion. The system keeps pulling outdated information—stuff from before 2021—instead of what was published yesterday. For my work, that's a dealbreaker. It's pushing me back to cloud LLMs, which defeats the purpose of going local.
Also the local text-to-speech did not work either. I can help bringing another yet more powerful algorithm to WebUI if that helps, but I think it is a front-end matter that the current tool is not functioning as intended.
@tjbck, for context: I'm a senior AI/CV researcher. My entire R&D cycle for conferences depends on knowing what has happened on the frontier of AI and CV e.g., on arXiv yesterday!. Current affairs are everything.
A powerful, self-hosted search feature isn't a nice-to-have; it's as critical as the core model. I've seen it done right with Perplexity and DeepSeek (which is awesome, by the way). My goal is to have that same capability, but here, under our own control.
I'm new to Open WebUI and not sure where to start contributing to make this happen, but I'm all in to help build it. This is a feature we desperately need. I'd be glad to change the world by my share one bit more! :)
<!-- gh-comment-id:3221868925 -->
@yekruz commented on GitHub (Aug 25, 2025):
Hey @rgaricano, thanks for the tip.
To be honest, I've hit a wall with your suggestion. The system keeps pulling outdated information—stuff from before 2021—instead of what was published yesterday. For my work, that's a dealbreaker. It's pushing me back to cloud LLMs, which defeats the purpose of going local.
Also the local text-to-speech did not work either. I can help bringing another yet more powerful algorithm to WebUI if that helps, but I think it is a front-end matter that the current tool is not functioning as intended.
@tjbck, for context: I'm a senior AI/CV researcher. My entire R&D cycle for conferences depends on knowing what has happened on the frontier of AI and CV e.g., on arXiv *yesterday*!. Current affairs are everything.
A powerful, self-hosted search feature isn't a nice-to-have; it's as critical as the core model. I've seen it done right with Perplexity and DeepSeek (which is awesome, by the way). My goal is to have that same capability, but here, under our own control.
I'm new to Open WebUI and not sure where to start contributing to make this happen, but I'm all in to help build it. This is a feature we desperately need. I'd be glad to change the world by my share one bit more! :)
@rgaricano
thanks got is in the first time. even with that, search does not yield any meaningful recent results, far away from those from early days of Perplexity.
how about the text-to-speech issue? do you also experience that?
<!-- gh-comment-id:3231555521 -->
@yekruz commented on GitHub (Aug 28, 2025):
@rgaricano
thanks got is in the first time. even with that, search does not yield any meaningful recent results, far away from those from early days of Perplexity.
how about the text-to-speech issue? do you also experience that?
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Originally created by @yekruz on GitHub (Aug 22, 2025).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/16804
Check Existing Issues
Installation Method
Docker
Open WebUI Version
v0.6.22
Ollama Version (if applicable)
0.11.5
Operating System
Ubuntu 24.04
Browser (if applicable)
Firefox 141.0.3
Confirmation
README.md.Expected Behavior
by selecting search button, I shall see that model can search through internet and has access to the real-time data on web similar to Perplexity or DeepSeek Search, ...
by clicking on the speaker icon below each reply, I shall hear the text turned into speech
Actual Behavior
I have tried with DuckDuckGo and TavilyAI and both failed.
I have a PhD in AI and can help to provide algorithms and their codes for something as similar as possible to the one from Perplexity and also Text-to-Speech, but I do not have background in website and UI development.
Searching is also very slow.
Steps to Reproduce
docker-composeeepseek-r1:32orgpt-oss:205-1) configure the search setting and activate search and do a query
5-2) click on the speaker icon below each reply
Logs & Screenshots
logs_docker_ollama-webui.txt
Additional Information
No response
@rgaricano commented on GitHub (Aug 22, 2025):
To fix the datetime issue of the model, try adding to system prompt:
for search engines
duckduckgo search engine have issues with concurrent request in some of the search engines that it use, it's improved in last version (now appear as DDGS) to reflect the max concurrent request value,
for tabily, as I know, it use max concurrent request, maybe you can do tests with low values to check it funcionality.
@yekruz commented on GitHub (Aug 25, 2025):
Hey @rgaricano, thanks for the tip.
To be honest, I've hit a wall with your suggestion. The system keeps pulling outdated information—stuff from before 2021—instead of what was published yesterday. For my work, that's a dealbreaker. It's pushing me back to cloud LLMs, which defeats the purpose of going local.
Also the local text-to-speech did not work either. I can help bringing another yet more powerful algorithm to WebUI if that helps, but I think it is a front-end matter that the current tool is not functioning as intended.
@tjbck, for context: I'm a senior AI/CV researcher. My entire R&D cycle for conferences depends on knowing what has happened on the frontier of AI and CV e.g., on arXiv yesterday!. Current affairs are everything.
A powerful, self-hosted search feature isn't a nice-to-have; it's as critical as the core model. I've seen it done right with Perplexity and DeepSeek (which is awesome, by the way). My goal is to have that same capability, but here, under our own control.
I'm new to Open WebUI and not sure where to start contributing to make this happen, but I'm all in to help build it. This is a feature we desperately need. I'd be glad to change the world by my share one bit more! :)
@rgaricano commented on GitHub (Aug 26, 2025):
e.g.
@yekruz commented on GitHub (Aug 28, 2025):
@rgaricano
thanks got is in the first time. even with that, search does not yield any meaningful recent results, far away from those from early days of Perplexity.
how about the text-to-speech issue? do you also experience that?