[GH-ISSUE #11736] feat: Separate image processing model #31866

Closed
opened 2026-04-25 05:45:32 -05:00 by GiteaMirror · 3 comments
Owner

Originally created by @huanyan434 on GitHub (Mar 16, 2025).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/11736

Check Existing Issues

  • I have searched the existing issues and discussions.

Problem Description

my model does not support image processing

Desired Solution you'd like

I want image processing to be done using an external model like web search, because my model does not support image processing

Alternatives Considered

No response

Additional Context

No response

Originally created by @huanyan434 on GitHub (Mar 16, 2025). Original GitHub issue: https://github.com/open-webui/open-webui/issues/11736 ### Check Existing Issues - [x] I have searched the existing issues and discussions. ### Problem Description my model does not support image processing ### Desired Solution you'd like I want image processing to be done using an external model like web search, because my model does not support image processing ### Alternatives Considered _No response_ ### Additional Context _No response_
Author
Owner

@Classic298 commented on GitHub (Mar 16, 2025):

  1. use a model that supports image processing and let it process the image
  2. switch back to the desired model (that doesn't support image processing) for the next response
<!-- gh-comment-id:2727372106 --> @Classic298 commented on GitHub (Mar 16, 2025): 1) use a model that supports image processing and let it process the image 2) switch back to the desired model (that doesn't support image processing) for the next response
Author
Owner

@spammenotinoz commented on GitHub (Mar 16, 2025):

Agree, for testing I hard-coded a Model when image gen is used, faster, costs less and less ramblings.
The real question is why even use an LLM at all?
Except for Gemini, where it understands and can make changes to the picture created.

<!-- gh-comment-id:2727686422 --> @spammenotinoz commented on GitHub (Mar 16, 2025): Agree, for testing I hard-coded a Model when image gen is used, faster, costs less and less ramblings. The real question is why even use an LLM at all? Except for Gemini, where it understands and can make changes to the picture created.
Author
Owner

@AbdullahMPrograms commented on GitHub (Mar 16, 2025):

this is doable using pipes, here is a rough one I created and have been using for the last while: https://github.com/AbdullahMPrograms/DynamicVisionRouter

<!-- gh-comment-id:2727705980 --> @AbdullahMPrograms commented on GitHub (Mar 16, 2025): this is doable using pipes, here is a rough one I created and have been using for the last while: https://github.com/AbdullahMPrograms/DynamicVisionRouter
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/open-webui#31866