github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 00:22:43 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #12851] Qwen3-VL does not support images (0.12.7)? #70575

New Issue
Open
opened 2026-05-04 22:03:26 -05:00 by GiteaMirror · 15 comments
GiteaMirror commented 2026-05-04 22:03:26 -05:00
Owner
Copy Link

Originally created by @ansorre on GitHub (Oct 30, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/12851

Originally assigned to: @hoyyeva on GitHub.

What is the issue?

As per app.log I'm on "Ollama version=0.12.7 OS=Windows/10.0.26100"
Using "qwen3-vl:30b" which as per info on https://ollama.com/library/qwen3-vl is able to accept "Text, Image" inputs.
Still I get the error: "The model does not support images" when I drop an image in the Ollama chat.
Is this the expected behavior? And if yes, why?

Relevant log output

OS

No response

GPU

No response

CPU

No response

Ollama version

No response

Originally created by @ansorre on GitHub (Oct 30, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/12851 Originally assigned to: @hoyyeva on GitHub. ### What is the issue? As per app.log I'm on "Ollama version=0.12.7 OS=Windows/10.0.26100" Using "qwen3-vl:30b" which as per info on https://ollama.com/library/qwen3-vl is able to accept "Text, Image" inputs. Still I get the error: "The model does not support images" when I drop an image in the Ollama chat. Is this the expected behavior? And if yes, why? ### Relevant log output ```shell ``` ### OS _No response_ ### GPU _No response_ ### CPU _No response_ ### Ollama version _No response_
GiteaMirror added the appbug labels 2026-05-04 22:03:27 -05:00
GiteaMirror commented 2026-05-04 22:03:29 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Oct 30, 2025):

Does it work from the command line?

$ ollama run qwen3-vl:30b describe this image: ./picture.png 
Added image './picture.png'
Thinking...
So, let's see. The image shows a small white puppy. 
...
...done thinking.

The image features a **small, fluffy white puppy** as the central subject, sitting 
...
<!-- gh-comment-id:3467094757 --> @rick-github commented on GitHub (Oct 30, 2025): Does it work from the command line? ```console $ ollama run qwen3-vl:30b describe this image: ./picture.png Added image './picture.png' Thinking... So, let's see. The image shows a small white puppy. ... ...done thinking. The image features a **small, fluffy white puppy** as the central subject, sitting ... ```
GiteaMirror commented 2026-05-04 22:03:29 -05:00
Author
Owner
Copy Link

@ansorre commented on GitHub (Oct 30, 2025):

@rick-github Yes, It works perfectly via command line! Sorry for not trying this before.

<!-- gh-comment-id:3467521483 --> @ansorre commented on GitHub (Oct 30, 2025): @rick-github Yes, It works perfectly via command line! Sorry for not trying this before.
GiteaMirror commented 2026-05-04 22:03:30 -05:00
Author
Owner
Copy Link

@rod-fu commented on GitHub (Oct 31, 2025):

i have a question that i wanna use python to call ollama api, can i do it on 0.12.7?

<!-- gh-comment-id:3471112334 --> @rod-fu commented on GitHub (Oct 31, 2025): i have a question that i wanna use python to call ollama api, can i do it on 0.12.7?
GiteaMirror commented 2026-05-04 22:03:30 -05:00
Author
Owner
Copy Link

@ansorre commented on GitHub (Oct 31, 2025):

i have a question that i wanna use python to call ollama api, can i do it on 0.12.7?

Yes, you definitely can. Ask ChatGPT/Claude/Gemini for help.

<!-- gh-comment-id:3473209542 --> @ansorre commented on GitHub (Oct 31, 2025): > i have a question that i wanna use python to call ollama api, can i do it on 0.12.7? Yes, you definitely can. Ask ChatGPT/Claude/Gemini for help.
GiteaMirror commented 2026-05-04 22:03:30 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Oct 31, 2025):

https://github.com/ollama/ollama-python

<!-- gh-comment-id:3473668189 --> @rick-github commented on GitHub (Oct 31, 2025): https://github.com/ollama/ollama-python
GiteaMirror commented 2026-05-04 22:03:31 -05:00
Author
Owner
Copy Link

@lisngt commented on GitHub (Nov 7, 2025):

Image

I have the same error.

<!-- gh-comment-id:3502690353 --> @lisngt commented on GitHub (Nov 7, 2025): <img width="1006" height="438" alt="Image" src="https://github.com/user-attachments/assets/21aeddb3-876f-42d0-a608-41374d24f809" /> I have the same error.
GiteaMirror commented 2026-05-04 22:03:32 -05:00
Author
Owner
Copy Link

@mistermantas commented on GitHub (Nov 9, 2025):

its a bug, i have the same issue too, cli is fine

<!-- gh-comment-id:3508280310 --> @mistermantas commented on GitHub (Nov 9, 2025): its a bug, i have the same issue too, cli is fine
GiteaMirror commented 2026-05-04 22:03:32 -05:00
Author
Owner
Copy Link

@GitOguz commented on GitHub (Nov 15, 2025):

same issue, cant let it do anything almost on a image.

<!-- gh-comment-id:3536966903 --> @GitOguz commented on GitHub (Nov 15, 2025): same issue, cant let it do anything almost on a image.
GiteaMirror commented 2026-05-04 22:03:33 -05:00
Author
Owner
Copy Link

@hoyyeva commented on GitHub (Nov 19, 2025):

Hello everyone! We are currently working on a fix. In the meantime, as a temporarily solution if you can try running ollama serve in a different terminal and then restart the app it should work. We are working on a more robust solution. Sorry for the inconvenience, and we will update here once the fix is ready!

<!-- gh-comment-id:3553913226 --> @hoyyeva commented on GitHub (Nov 19, 2025): Hello everyone! We are currently working on a fix. In the meantime, as a temporarily solution if you can try running `ollama serve` in a different terminal and then restart the app it should work. We are working on a more robust solution. Sorry for the inconvenience, and we will update here once the fix is ready!
GiteaMirror commented 2026-05-04 22:03:33 -05:00
Author
Owner
Copy Link

@theodoruszq commented on GitHub (Dec 9, 2025):

Hello everyone! We are currently working on a fix. In the meantime, as a temporarily solution if you can try running ollama serve in a different terminal and then restart the app it should work. We are working on a more robust solution. Sorry for the inconvenience, and we will update here once the fix is ready!

hi could you please check this model?

Image

I tried qwen3-vl:4b etc with the lastest ollama(0.13.2), the pasted image works well, but not with this qwen3-vl:30b/qwen3-vl:8b, they failed.

By the way, could you give some suggestions to disable thinking for qwen-vl model? I tried add special token to my prompt, but it doesn't work.

<!-- gh-comment-id:3631474630 --> @theodoruszq commented on GitHub (Dec 9, 2025): > Hello everyone! We are currently working on a fix. In the meantime, as a temporarily solution if you can try running `ollama serve` in a different terminal and then restart the app it should work. We are working on a more robust solution. Sorry for the inconvenience, and we will update here once the fix is ready! hi could you please check this model? <img width="681" height="155" alt="Image" src="https://github.com/user-attachments/assets/fb102159-ecfa-49a5-b404-1008cea5e8d9" /> I tried qwen3-vl:4b etc with the lastest ollama(0.13.2), the pasted image works well, but not with this qwen3-vl:30b/qwen3-vl:8b, they failed. By the way, could you give some suggestions to disable thinking for qwen-vl model? I tried add </think> special token to my prompt, but it doesn't work.
GiteaMirror commented 2026-05-04 22:03:34 -05:00
Author
Owner
Copy Link

@teddybear082 commented on GitHub (Dec 13, 2025):

By the way, could you give some suggestions to disable thinking for qwen-vl model? I tried add special token to my prompt, but it doesn't work.

Use the instruct version of the models. When you go to the ollama models web page click the button for more models and you will see them. Instruct versions won’t do the thinking stuff.

<!-- gh-comment-id:3649563391 --> @teddybear082 commented on GitHub (Dec 13, 2025): > > > > By the way, could you give some suggestions to disable thinking for qwen-vl model? I tried add special token to my prompt, but it doesn't work. Use the instruct version of the models. When you go to the ollama models web page click the button for more models and you will see them. Instruct versions won’t do the thinking stuff.
GiteaMirror commented 2026-05-04 22:03:35 -05:00
Author
Owner
Copy Link

@hyacin75 commented on GitHub (Dec 15, 2025):

Hello everyone! We are currently working on a fix. In the meantime, as a temporarily solution if you can try running ollama serve in a different terminal and then restart the app it should work. We are working on a more robust solution. Sorry for the inconvenience, and we will update here once the fix is ready!

Not working for me - perhaps I'm doing it wrong? Shut down GUI app, run 'ollama serve' in a terminal, re-launch GUI app? Still seeing the same 'not supported' messages.

<!-- gh-comment-id:3656752416 --> @hyacin75 commented on GitHub (Dec 15, 2025): > Hello everyone! We are currently working on a fix. In the meantime, as a temporarily solution if you can try running `ollama serve` in a different terminal and then restart the app it should work. We are working on a more robust solution. Sorry for the inconvenience, and we will update here once the fix is ready! Not working for me - perhaps I'm doing it wrong? Shut down GUI app, run 'ollama serve' in a terminal, re-launch GUI app? Still seeing the same 'not supported' messages.
GiteaMirror commented 2026-05-04 22:03:35 -05:00
Author
Owner
Copy Link

@hoyyeva commented on GitHub (Dec 15, 2025):

Hi @theodoruszq & @hyacin75 do you have the model downloaded already? We noticed that there is a limitation on one of the ollama endpoint to get correct capabilities when the model is not downloaded. We are currently working on an improved ollama API for model search which shoul address this issue as well. For now, the workaround is to download the model first by starting a chat without attaching an image. Once the download is complete, you should be able to attach image files.

<!-- gh-comment-id:3657515842 --> @hoyyeva commented on GitHub (Dec 15, 2025): Hi @theodoruszq & @hyacin75 do you have the model downloaded already? We noticed that there is a limitation on one of the ollama endpoint to get correct capabilities when the model is not downloaded. We are currently working on an improved ollama API for model search which shoul address this issue as well. For now, the workaround is to download the model first by starting a chat without attaching an image. Once the download is complete, you should be able to attach image files.
GiteaMirror commented 2026-05-04 22:03:36 -05:00
Author
Owner
Copy Link

@hyacin75 commented on GitHub (Dec 15, 2025):

@hoyyeva Possible I did not - I gave up trying and just fired it up on the CLI, and iirc it downloaded the model first thing. I was able to load the images I wanted analyzed though and got what I needed thanks to all the great tips and workaround here! :-) It was just a one-off for me to get a prompt I could feed back into the non-vision model, to have it spit out an SD prompt, lol, AI helping me guide AI to control another AI... head spinning. 😵‍💫 🤣

<!-- gh-comment-id:3657559075 --> @hyacin75 commented on GitHub (Dec 15, 2025): @hoyyeva Possible I did not - I gave up trying and just fired it up on the CLI, and iirc it downloaded the model first thing. I was able to load the images I wanted analyzed though and got what I needed thanks to all the great tips and workaround here! :-) It was just a one-off for me to get a prompt I could feed back into the non-vision model, to have it spit out an SD prompt, lol, AI helping me guide AI to control another AI... head spinning. 😵‍💫 🤣
GiteaMirror commented 2026-05-04 22:03:36 -05:00
Author
Owner
Copy Link

@theodoruszq commented on GitHub (Dec 24, 2025):

Hi @theodoruszq & @hyacin75 do you have the model downloaded already? We noticed that there is a limitation on one of the ollama endpoint to get correct capabilities when the model is not downloaded. We are currently working on an improved ollama API for model search which shoul address this issue as well. For now, the workaround is to download the model first by starting a chat without attaching an image. Once the download is complete, you should be able to attach image files.

Actually I have the model downloaded already, it maybe a bug. I am wondering some easy python code to debugging with... Finally I switched to GPT-OSS20B to solve my offline request.

Thanks anyway~

<!-- gh-comment-id:3690221243 --> @theodoruszq commented on GitHub (Dec 24, 2025): > Hi [@theodoruszq](https://github.com/theodoruszq) & [@hyacin75](https://github.com/hyacin75) do you have the model downloaded already? We noticed that there is a limitation on one of the ollama endpoint to get correct capabilities when the model is not downloaded. We are currently working on an improved ollama API for model search which shoul address this issue as well. For now, the workaround is to download the model first by starting a chat without attaching an image. Once the download is complete, you should be able to attach image files. Actually I have the model downloaded already, it maybe a bug. I am wondering some easy python code to debugging with... Finally I switched to GPT-OSS20B to solve my offline request. Thanks anyway~
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label app bug
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#70575
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?