[GH-ISSUE #24375] issue: Image editing fails in Edge but works in Chrome #58950

New Issue

GiteaMirror · 2026-05-06T00:35:59-05:00

GiteaMirror commented

2026-05-06 00:35:59 -05:00

Originally created by @ambrustorok on GitHub (May 5, 2026).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/24375

Check Existing Issues

I have searched for any existing and/or related issues.
I have searched for any existing and/or related discussions.
I have also searched in the CLOSED issues AND CLOSED discussions and found no related items (your issue might already be addressed on the development branch!).
I am using the latest version of Open WebUI.

Installation Method

Other

Open WebUI Version

v0.9.2

Ollama Version (if applicable)

No response

Operating System

Windows 11

Browser (if applicable)

Edge / Chrome

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

When an image is uploaded in Open WebUI and then used with an image-editing capable model/tool, the uploaded image should be passed to the edit_image tool in a form that the tool can access, regardless of browser.

Specifically, Microsoft Edge should behave the same as Google Chrome:

Upload a JPG or PNG image.
Send a prompt requesting image editing.
The image-editing tool receives an accessible image URL or file reference.
The edited image is generated successfully.

Actual Behavior

Image editing consistently fails when using Microsoft Edge, but succeeds in Google Chrome with the same Open WebUI instance, same account, same model, same prompt, and same uploaded image file.

The issue appears to be that, in Edge, the uploaded image is not passed to the edit_image tool as an accessible Open WebUI file URL. Instead, the tool sometimes receives only a plain filename, for example:

image_urls: ["chair.png"]

or:

image_urls: ["bikes.jpg"]

In other cases, Edge appears to pass or hallucinate a local/non-accessible path, for example:

image_urls: ["/mnt/data/0.png"]

These references are not accessible to the image-editing tool, so the tool fails with:

{
  "error": "400: [ERROR: 404: We could not find what you're looking for :/]"
}

The assistant then reports that it could not access the uploaded image file.

In Chrome, the same workflow succeeds. The edit_image tool receives an accessible internal Open WebUI file URL, for example:

<file-id>

and the image edit is generated successfully.

This happens with both PNG and JPG images. It also appears to be model-independent: the same browser-specific behavior occurs with at least:

gpt-5.5
Claude Sonnet 4.6

In Edge, the image reference can become a plain filename or a hallucinated/non-accessible local path instead of the expected /api/v1/files/<file-id>/content URL.

Steps to Reproduce

Open OpenWebUI in Microsoft Edge on Windows.
Start a new chat and upload an image, e.g. chair.png.
Send an image-editing prompt, e.g.
```
replace chair with a coatrack
```
The image is shown in the chat, but the edit_image tool fails.
Expanding the tool details shows that Edge passes only a filename or non-accessible path, e.g.
```
image_urls: ["chair.png"]
```
or:
```
image_urls: ["/mnt/data/0.png"]
```

The tool returns:

{
  "error": "400: [ERROR: 404: We could not find what you're looking for :/]"
}

Repeat the same steps in Google Chrome with the same Open WebUI instance, account, model, image, and prompt.
In Chrome, the edit succeeds, and the tool receives an accessible file URL, e.g.
```
<file-id>
```
The same pattern occurs with both PNG and JPG images.

Logs & Screenshots

PNG images

JPG images

hallucinated path

Additional Information

img sources:

Originally created by @ambrustorok on GitHub (May 5, 2026). Original GitHub issue: https://github.com/open-webui/open-webui/issues/24375 ### Check Existing Issues - [x] I have searched for any existing and/or related issues. - [x] I have searched for any existing and/or related discussions. - [x] I have also searched in the CLOSED issues AND CLOSED discussions and found no related items (your issue might already be addressed on the development branch!). - [x] I am using the latest version of Open WebUI. ### Installation Method Other ### Open WebUI Version v0.9.2 ### Ollama Version (if applicable) _No response_ ### Operating System Windows 11 ### Browser (if applicable) Edge / Chrome ### 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. - [ ] 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). - [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 When an image is uploaded in Open WebUI and then used with an image-editing capable model/tool, the uploaded image should be passed to the `edit_image` tool in a form that the tool can access, regardless of browser. Specifically, Microsoft Edge should behave the same as Google Chrome: 1. Upload a JPG or PNG image. 2. Send a prompt requesting image editing. 3. The image-editing tool receives an accessible image URL or file reference. 4. The edited image is generated successfully. ### Actual Behavior Image editing consistently fails when using Microsoft Edge, but succeeds in Google Chrome with the same Open WebUI instance, same account, same model, same prompt, and same uploaded image file. The issue appears to be that, in Edge, the uploaded image is not passed to the `edit_image` tool as an accessible Open WebUI file URL. Instead, the tool sometimes receives only a plain filename, for example: ```text image_urls: ["chair.png"] ``` or: ```text image_urls: ["bikes.jpg"] ``` In other cases, Edge appears to pass or hallucinate a local/non-accessible path, for example: ```text image_urls: ["/mnt/data/0.png"] ``` These references are not accessible to the image-editing tool, so the tool fails with: ```json { "error": "400: [ERROR: 404: We could not find what you're looking for :/]" } ``` The assistant then reports that it could not access the uploaded image file. In Chrome, the same workflow succeeds. The `edit_image` tool receives an accessible internal Open WebUI file URL, for example: ```text <file-id> ``` and the image edit is generated successfully. This happens with both PNG and JPG images. It also appears to be model-independent: the same browser-specific behavior occurs with at least: - `gpt-5.5` - `Claude Sonnet 4.6` In Edge, the image reference can become a plain filename or a hallucinated/non-accessible local path instead of the expected `/api/v1/files/<file-id>/content` URL. ### Steps to Reproduce 1. Open OpenWebUI in Microsoft Edge on Windows. 2. Start a new chat and upload an image, e.g. `chair.png`. 3. Send an image-editing prompt, e.g. ```text replace chair with a coatrack ``` 4. The image is shown in the chat, but the `edit_image` tool fails. 5. Expanding the tool details shows that Edge passes only a filename or non-accessible path, e.g. ```text image_urls: ["chair.png"] ``` or: ```text image_urls: ["/mnt/data/0.png"] ``` 6. The tool returns: ```json { "error": "400: [ERROR: 404: We could not find what you're looking for :/]" } ``` 7. Repeat the same steps in Google Chrome with the same Open WebUI instance, account, model, image, and prompt. 8. In Chrome, the edit succeeds, and the tool receives an accessible file URL, e.g. ```text <file-id> ``` 9. The same pattern occurs with both PNG and JPG images. ### Logs & Screenshots ### PNG images <img width="1440" height="1999" alt="Image" src="https://github.com/user-attachments/assets/daa3428d-db26-48f0-a0a6-5bd1921fdb95" /> ### JPG images <img width="1436" height="2281" alt="Image" src="https://github.com/user-attachments/assets/444eb056-0eba-4975-81d0-e6aee58c20af" /> ### hallucinated path <img width="966" height="693" alt="Image" src="https://github.com/user-attachments/assets/cbe1873e-b34c-4d79-b79a-32457dcb46a0" /> ### Additional Information img sources: - https://www.pexels.com/photo/charming-cafe-exterior-in-h-chi-minh-city-33971235/ - https://www.pexels.com/photo/outdoor-cafe-gathering-under-a-tree-34743999/

GiteaMirror added the bug label 2026-05-06 00:35:59 -05:00

GiteaMirror commented

2026-05-06 00:36:05 -05:00

@owui-terminator[bot] commented on GitHub (May 5, 2026):

🔍 Similar Issues Found

I found some existing issues that might be related. Please check if any of these are duplicates or contain helpful solutions:

#20754 issue: New upload of an image does not work with image editing
by iChristGit · bug
#21114 issue: Multiple image edits do not meet expectations.
by gitFox117 · bug
#20237 issue: Image editing stopped working
by tomasloksa · bug, confirmed issue
#20490 issue: Unable to edit reference image
by fgonzalez-glmc · bug
#19083 issue: output always copied with formatting (chrome)
by tomasloksa · bug, good first issue

💡 If this is a duplicate, consider closing it and adding details to the existing issue.

This comment was generated automatically. React with 👍 if helpful, 👎 if not.

@owui-terminator[bot] commented on GitHub (May 5, 2026): 🔍 **Similar Issues Found** I found some existing issues that might be related. Please check if any of these are duplicates or contain helpful solutions: 1. [#20754](https://github.com/open-webui/open-webui/issues/20754) **issue: New upload of an image does not work with image editing** *by iChristGit · `bug`* 2. [#21114](https://github.com/open-webui/open-webui/issues/21114) **issue: Multiple image edits do not meet expectations.** *by gitFox117 · `bug`* 3. [#20237](https://github.com/open-webui/open-webui/issues/20237) **issue: Image editing stopped working** *by tomasloksa · `bug`, `confirmed issue`* 4. [#20490](https://github.com/open-webui/open-webui/issues/20490) **issue: Unable to edit reference image** *by fgonzalez-glmc · `bug`* 5. [#19083](https://github.com/open-webui/open-webui/issues/19083) **issue: output always copied with formatting (chrome)** *by tomasloksa · `bug`, `good first issue`* --- 💡 If this is a duplicate, consider closing it and adding details to the existing issue. *This comment was generated automatically.* React with 👍 if helpful, 👎 if not.

GiteaMirror commented

2026-05-06 00:36:09 -05:00

@ambrustorok commented on GitHub (May 5, 2026):

The similar issues have all been closed and there has been a release since, thus I deemed it best to open a new one.

@ambrustorok commented on GitHub (May 5, 2026): The similar issues have all been closed and there has been a release since, thus I deemed it best to open a new one.

GiteaMirror commented

2026-05-06 00:36:13 -05:00

@Classic298 commented on GitHub (May 5, 2026):

I cannot reproduce in Edge, at all - but keep us updated. I am sure this has to be some Edge specific security setting similar to how we found out this: https://docs.openwebui.com/features/chat-conversations/chat-features/code-execution/python#microsoft-edge-pyodide-crashes

Let us know!

@Classic298 commented on GitHub (May 5, 2026): I cannot reproduce in Edge, at all - but keep us updated. I am sure this has to be some Edge specific security setting similar to how we found out this: https://docs.openwebui.com/features/chat-conversations/chat-features/code-execution/python#microsoft-edge-pyodide-crashes Let us know!

Sign in to join this conversation.

Branches Tags

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: github-starred/open-webui#58950