[GH-ISSUE #15579] issue: Ask/Explain buttons in chat context do not use users OpenAI direct connections #17608

New Issue

GiteaMirror · 2026-04-19T23:24:39-05:00

GiteaMirror commented

2026-04-19 23:24:39 -05:00

Originally created by @wilstdu on GitHub (Jul 8, 2025).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/15579

Check Existing Issues

I have searched the existing issues and discussions.
I am using the latest version of Open WebUI.

Installation Method

Docker

Open WebUI Version

0.6.14

Ollama Version (if applicable)

No response

Operating System

macOS, Ubuntu

Browser (if applicable)

No response

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

If a user has a direct connection to OpenAI-compatible service configured, it should also use that connection settings when trying to fetch any OpenAI format information.

In my case, it was only about the Ask and Explain buttons, but there should be more places where this is a problem.

Actual Behavior

Our setup:

LiteLLM
OpenWebUI (0.6.14 in the cluster, same problems when started locally with the latest version 0.6.15)
Kubernetes cluster

Users create their tokens in LiteLLM and configure them in OpenWebUI - Settings - Connections - Manage Direct Connections:

URL - LiteLLM domain address
Key - LiteLLM user API key

Behaviour:
When trying to continue a discussion about certain points of the generated response in the chat via these buttons:

It results an error:

After setting these environment variables:
OPENAI_API_BASE_URLS
OPENAI_API_KEYS
It works.

The problem with those variables is that they are set system-wide, meaning that it would need a technical user in LiteLLM, and the requests that the individual user does would go to the technical user`s bills (doesn't make sense, heh?).

Steps to Reproduce

Make sure that OPENAI_API_BASE_URLS and OPENAI_API_KEYS are not set
Start a fresh local instance of OpenWebUI - it's reproducible everywhere - Docker, pip install, local dev.sh script
Setup the admin user if needed, and login
Go to OpenWebUI - Settings - Connections - Manage Direct Connections:
URL - your OpenAI domain address
Key - your OpenAI API key
Start a chat, ask anything, wait for response
After getting the response - double click on any text from the response
Click Ask or Explain - an error will show up after sending the request

Logs & Screenshots

OpenWebUI logs: openwebui-problem.log

Additional Information

This is the location where the error is thrown:

The main root cause:

Originally created by @wilstdu on GitHub (Jul 8, 2025). Original GitHub issue: https://github.com/open-webui/open-webui/issues/15579 ### 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 0.6.14 ### Ollama Version (if applicable) _No response_ ### Operating System macOS, Ubuntu ### Browser (if applicable) _No response_ ### 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 If a user has a direct connection to OpenAI-compatible service configured, it should also use that connection settings when trying to fetch any OpenAI format information. In my case, it was only about the `Ask` and `Explain` buttons, but there should be more places where this is a problem. ### Actual Behavior Our setup: - LiteLLM - OpenWebUI (0.6.14 in the cluster, same problems when started locally with the latest version 0.6.15) - Kubernetes cluster Users create their tokens in LiteLLM and configure them in OpenWebUI - Settings - Connections - Manage Direct Connections: - URL - LiteLLM domain address - Key - LiteLLM user API key Behaviour: When trying to continue a discussion about certain points of the generated response in the chat via these buttons: <img width="1087" height="367" alt="Image" src="https://github.com/user-attachments/assets/65f65008-0ab6-4806-9811-d83843e810b7" /> It results an error: <img width="2264" height="515" alt="Image" src="https://github.com/user-attachments/assets/898fd002-9e2d-41d5-b576-b671f5902013" /> After setting these environment variables: `OPENAI_API_BASE_URLS` `OPENAI_API_KEYS` It works. The problem with those variables is that they are set system-wide, meaning that it would need a technical user in LiteLLM, and the requests that the individual user does would go to the technical user`s bills (doesn't make sense, heh?). ### Steps to Reproduce 1. Make sure that OPENAI_API_BASE_URLS and OPENAI_API_KEYS are not set 2. Start a fresh local instance of OpenWebUI - it's reproducible everywhere - Docker, pip install, local dev.sh script 3. Setup the admin user if needed, and login 4. Go to OpenWebUI - Settings - Connections - Manage Direct Connections: URL - your OpenAI domain address Key - your OpenAI API key 5. Start a chat, ask anything, wait for response 6. After getting the response - double click on any text from the response 7. Click Ask or Explain - an error will show up after sending the request ### Logs & Screenshots OpenWebUI logs: [openwebui-problem.log](https://github.com/user-attachments/files/21116679/openwebui-problem.log) ### Additional Information This is the location where the error is thrown: <img width="784" height="385" alt="Image" src="https://github.com/user-attachments/assets/d89ed562-5405-4ed5-b225-a0075b19c62f" /> The main root cause: <img width="1152" height="566" alt="Image" src="https://github.com/user-attachments/assets/0a3f5ed9-2327-44e6-8c22-b1c9ac9a011a" />

GiteaMirror added the bug label 2026-04-19 23:24:39 -05:00

GiteaMirror referenced this issue

2026-04-20 05:25:37 -05:00

[PR #17608] [MERGED] fix: prevent memory leaks in file handling and HTTP connections #24486

GiteaMirror referenced this issue

2026-04-25 12:31:57 -05:00

[PR #17608] [MERGED] fix: prevent memory leaks in file handling and HTTP connections #40116

GiteaMirror referenced this issue

2026-04-29 22:48:46 -05:00

[PR #17608] [MERGED] fix: prevent memory leaks in file handling and HTTP connections #47534

GiteaMirror referenced this issue

2026-05-06 08:03:56 -05:00