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 #14782] Allow Control List for AI Models Downloaded #9552

New Issue
Closed
opened 2026-04-12 22:28:17 -05:00 by GiteaMirror · 3 comments
GiteaMirror commented 2026-04-12 22:28:17 -05:00
Owner
Copy Link

Originally created by @jamboNum5 on GitHub (Mar 11, 2026).
Original GitHub issue: https://github.com/ollama/ollama/issues/14782

I work as a sys-admin at a University and there is a growing demand for using Ollama in computer labs for teaching. There are a few considerations we have for faciliating AI models in Labs.

  • Due to data governance policies, we have to consider what Student data is uploaded to cloud services. This limits the kind of models we use on local machines.
  • We need to ensure we are working inline with the licence terms of a given AI model, the licences seem to vary between models and we don't have the ability to control what model is downloaded.
  • To optimise performance, we would likely want to have the models installed locally on the machine rather than in user profiles.

This feels like a big ask and is more than one feature, but I was keen to better understand whether the following was a consideration for development:

  • Models stored on local disks rather than user profile (located on network shares). This would optimise read/write locally but also mean models aren't being downloaded from servers by individual users, reducing duplicating data and saving bandwidth.
  • ACL type feature to limit which models are downloaded and limit to approved models.

Thanks in advance

Originally created by @jamboNum5 on GitHub (Mar 11, 2026). Original GitHub issue: https://github.com/ollama/ollama/issues/14782 I work as a sys-admin at a University and there is a growing demand for using Ollama in computer labs for teaching. There are a few considerations we have for faciliating AI models in Labs. - Due to data governance policies, we have to consider what Student data is uploaded to cloud services. This limits the kind of models we use on local machines. - We need to ensure we are working inline with the licence terms of a given AI model, the licences seem to vary between models and we don't have the ability to control what model is downloaded. - To optimise performance, we would likely want to have the models installed locally on the machine rather than in user profiles. This feels like a big ask and is more than one feature, but I was keen to better understand whether the following was a consideration for development: - Models stored on local disks rather than user profile (located on network shares). This would optimise read/write locally but also mean models aren't being downloaded from servers by individual users, reducing duplicating data and saving bandwidth. - ACL type feature to limit which models are downloaded and limit to approved models. Thanks in advance
GiteaMirror added the feature request label 2026-04-12 22:28:17 -05:00
GiteaMirror commented 2026-04-12 22:28:19 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Mar 11, 2026):

https://github.com/ollama/ollama/issues/11667#issuecomment-3155719517

Control is easier if a single ollama instance is available to the students (as opposed to them running their own ollama servers). You can restrict access to just the OpenAI compatibility endpoint, this allows them to use the available models but they cannot download new ones. It does mean that they would have to request a new model from the sys-admin department if it was pertinent to their studies.

Storing models on local disk only makes sense if the students are running their own servers, in which case you would need to firewall off ollama.ai, hf.co, modelscope.cn etc to prevent them from downloading random models.

<!-- gh-comment-id:4039797550 --> @rick-github commented on GitHub (Mar 11, 2026): https://github.com/ollama/ollama/issues/11667#issuecomment-3155719517 Control is easier if a single ollama instance is available to the students (as opposed to them running their own ollama servers). You can restrict access to just the OpenAI compatibility endpoint, this allows them to use the available models but they cannot download new ones. It does mean that they would have to request a new model from the sys-admin department if it was pertinent to their studies. Storing models on local disk only makes sense if the students are running their own servers, in which case you would need to firewall off ollama.ai, hf.co, modelscope.cn etc to prevent them from downloading random models.
GiteaMirror commented 2026-04-12 22:28:19 -05:00
Author
Owner
Copy Link

@fcorneli commented on GitHub (Mar 11, 2026):

Simply put the allowed models somewhere on the machines, and give your students the path, which they can use as follows:

export OLLAMA_MODELS=/path/where/the/policy/allowed/ollama/models/live/
ollama ls

Of course they can always pull in something else I guess. Would have to see whether ollama can work on a read-only $OLLAMA_MODELS directory.

<!-- gh-comment-id:4039849830 --> @fcorneli commented on GitHub (Mar 11, 2026): Simply put the allowed models somewhere on the machines, and give your students the path, which they can use as follows: ``` export OLLAMA_MODELS=/path/where/the/policy/allowed/ollama/models/live/ ollama ls ``` Of course they can always pull in something else I guess. Would have to see whether `ollama` can work on a read-only `$OLLAMA_MODELS` directory.
GiteaMirror commented 2026-04-12 22:28:20 -05:00
Author
Owner
Copy Link

@jamboNum5 commented on GitHub (Mar 13, 2026):

Thank you both for the suggestions, between firewalling and the environment variable, that should be a great place to start.

I believe they are intending to run a local instance to interact with some robots. They wanted to give students more hands on experience configuring the models for work later on.

I'll set this feature to closed for the time being. Thanks again!

<!-- gh-comment-id:4053708952 --> @jamboNum5 commented on GitHub (Mar 13, 2026): Thank you both for the suggestions, between firewalling and the environment variable, that should be a great place to start. I believe they are intending to run a local instance to interact with some robots. They wanted to give students more hands on experience configuring the models for work later on. I'll set this feature to closed for the time being. Thanks again!
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 feature request
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#9552
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?