github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-07-16 21:25:46 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #11600] Ollama UI wants do download models that I already have #33418

New Issue
Open
opened 2026-04-22 16:04:03 -05:00 by GiteaMirror · 6 comments
GiteaMirror commented 2026-04-22 16:04:03 -05:00
Owner
Copy Link

Originally created by @philippgille on GitHub (Jul 31, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/11600

Originally assigned to: @BruceMacD on GitHub.

What is the issue?

After updating to the new Ollama version with a built-in UI, it shows the chat window with a pre-selected gemma3:4b model by default. That one, and others in the drop-down, all have to be downloaded first, see screenshot:

Image

But that doesn't make sense, as I already have some of those models (same hash, just with the more explicit quantization tag, like gemma3:4b-it-q4_K_M, see screenshot:

Image

=> I would consider the model detection buggy, as it should check for the hash, and then show gemma3:4b-it-q4_K_M by default, and similar for the others, e.g. stop showing qwen3:4b in the dropdown as qwen3:4b-q4_K_M already exists.

Relevant log output

/

OS

macOS Sequoia 15.5 arm64

GPU

Apple M1 Pro (14) @ 1.30 GHz [Integrated]

CPU

Apple M1 Pro (8) @ 3.23 GHz

Ollama version

$ ollama -v
ollama version is 0.10.1
Originally created by @philippgille on GitHub (Jul 31, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/11600 Originally assigned to: @BruceMacD on GitHub. ### What is the issue? After updating to the new Ollama version with a built-in UI, it shows the chat window with a pre-selected `gemma3:4b` model by default. That one, and others in the drop-down, all have to be downloaded first, see screenshot: <img width="410" height="402" alt="Image" src="https://github.com/user-attachments/assets/dc3e4915-e2d9-4b35-b693-c438f4b43ec4" /> But that doesn't make sense, as I already have some of those models (same hash, just with the more explicit quantization tag, like `gemma3:4b-it-q4_K_M`, see screenshot: <img width="412" height="287" alt="Image" src="https://github.com/user-attachments/assets/fa2004dc-7f78-4e38-a0ee-c8044a7889b2" /> => I would consider the model detection buggy, as it should check for the hash, and then show `gemma3:4b-it-q4_K_M` by default, and similar for the others, e.g. stop showing `qwen3:4b` in the dropdown as `qwen3:4b-q4_K_M` already exists. ### Relevant log output / ### OS macOS Sequoia 15.5 arm64 ### GPU Apple M1 Pro (14) @ 1.30 GHz [Integrated] ### CPU Apple M1 Pro (8) @ 3.23 GHz ### Ollama version ```console $ ollama -v ollama version is 0.10.1 ```
GiteaMirror added the bug label 2026-04-22 16:04:03 -05:00
GiteaMirror commented 2026-04-22 16:04:04 -05:00
Author
Owner
Copy Link

@talperets commented on GitHub (Jul 31, 2025):

Hey, I would love to fix it but I can't find the repository for the front end client. Is it even open source? I've already implemented a very similar feature in an ollama client that was made prior to the official app.

<!-- gh-comment-id:3139418454 --> @talperets commented on GitHub (Jul 31, 2025): Hey, I would love to fix it but I can't find the repository for the front end client. Is it even open source? I've already implemented a very similar feature in an ollama client that was made prior to the official app.
GiteaMirror commented 2026-04-22 16:04:05 -05:00
Author
Owner
Copy Link

@philippgille commented on GitHub (Jul 31, 2025):

Good point, I think it's closed source.

I'll keep the issue open as long as there's no other place to report this, and also I only ever installed the CLI version (from the GitHub releases here IIRC) and they just included the closed source UI with an automatic update, so it seems they consider it part of this repo's product.

<!-- gh-comment-id:3139513041 --> @philippgille commented on GitHub (Jul 31, 2025): Good point, I think it's closed source. I'll keep the issue open as long as there's no other place to report this, and also I only ever installed the CLI version (from the GitHub releases here IIRC) and they just included the closed source UI with an automatic update, so it seems they consider it part of this repo's product.
GiteaMirror commented 2026-04-22 16:04:05 -05:00
Author
Owner
Copy Link

@sammcj commented on GitHub (Jul 31, 2025):

I actually went to look at the code to see if I could fix this and submit a PR, then when I couldn't find the UI in the source I found this ticket and realised the latest version of the app doesn't seem to be open source anymore!

<!-- gh-comment-id:3139623337 --> @sammcj commented on GitHub (Jul 31, 2025): I actually went to look at the code to see if I could fix this and submit a PR, then when I couldn't find the UI in the source I found this ticket and realised the latest version of the app doesn't seem to be open source anymore!
GiteaMirror commented 2026-04-22 16:04:05 -05:00
Author
Owner
Copy Link

@talperets commented on GitHub (Jul 31, 2025):

That's very unfortunate

<!-- gh-comment-id:3139905938 --> @talperets commented on GitHub (Jul 31, 2025): That's very unfortunate
GiteaMirror commented 2026-04-22 16:04:06 -05:00
Author
Owner
Copy Link

@FellowTraveler commented on GitHub (Aug 2, 2025):

What I liked about Ollama was that it was open source and ran local models.

What I don't like: I already have the q8_0 model or the q6_K_M model, but the new UI downloads a q4_0 or q4_K_M by default.

I understand that most dipshit users need a heavily quantized model, and q4_K_M is what's best for them, and they don't know (or care) about the quantizations necessarily.

I understand the importance of having it "just work" for many users without them having to be super educated on the subject.

But the problem is, these models take up a lot of space on the SSD, and I don't want redundant downloads of models, I have already chosen the quants I want to use, and I want control / visibility over the quants I am using.

I also use Ollama as a local API for MANY other apps that access it. So I need to know what models/quants I am serving.

I think what is needed is a developer mode. That way Ollama can serve laymen and advanced users alike.

<!-- gh-comment-id:3146373475 --> @FellowTraveler commented on GitHub (Aug 2, 2025): What I liked about Ollama was that it was open source and ran local models. What I don't like: I already have the q8_0 model or the q6_K_M model, but the new UI downloads a q4_0 or q4_K_M by default. I understand that most dipshit users need a heavily quantized model, and q4_K_M is what's best for them, and they don't know (or care) about the quantizations necessarily. I understand the importance of having it "just work" for many users without them having to be super educated on the subject. But the problem is, these models take up a lot of space on the SSD, and I don't want redundant downloads of models, I have already chosen the quants I want to use, and I want control / visibility over the quants I am using. I also use Ollama as a local API for MANY other apps that access it. So I need to know what models/quants I am serving. **I think what is needed is a developer mode. That way Ollama can serve laymen and advanced users alike.**
GiteaMirror commented 2026-04-22 16:04:06 -05:00
Author
Owner
Copy Link

@JaneDoe84 commented on GitHub (Aug 5, 2025):

@FellowTraveler

What I liked about Ollama was that it was open source and ran local models.
I understand that most dipshit users need a heavily quantized model, and q4_K_M is what's best for them, and they don't know (or care) about the quantizations necessarily.

i do not understand that you vent your anger in this way !
im such a dipshit user, did not have the money to buy the stuff you seams to have.
but I have been programming for 30 years and am a supporter of open source, maybe even things YOU use.
It's a real shame that these days you can't exclude certain users from using your free work. You would be one of them.

<!-- gh-comment-id:3154786359 --> @JaneDoe84 commented on GitHub (Aug 5, 2025): @FellowTraveler > What I liked about Ollama was that it was open source and ran local models. > I understand that most **dipshit users** need a heavily quantized model, and q4_K_M is what's best for them, and they don't know (or care) about the quantizations necessarily. i do **not** understand that you vent your anger in this way ! im such a **dipshit** user, did not have the money to buy the stuff you seams to have. but I have been programming for 30 years and am a supporter of open source, maybe even things YOU use. It's a real shame that these days you can't exclude certain users from using your free work. You would be one of them.
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 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#33418
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?