github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-09 01:59:50 -05:00
Code Issues 14.6k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #2121] Feature request: control session duration of loaded models #1212

New Issue
Closed
opened 2026-04-12 10:59:14 -05:00 by GiteaMirror · 1 comment
GiteaMirror commented 2026-04-12 10:59:14 -05:00
Owner
Copy Link

Originally created by @nperez on GitHub (Jan 21, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/2121

I have a use case where multiple processes (stable diffusion, whsiper, ollama, etc) are competing for limited GPU resources and I need to share the GPU. Unfortunately, there doesn't appear to be a way to manage the session lifetime of loaded models in ollama. It would be cool to have the ability via model options to control the session lifetime (ie. unload after each request) or have a new endpoint to unconditionally unload whatever model is loaded. Without this feature, I need to manage (kill, then restart) the ollama process or wait the five minutes that is the current defaultSessionDuration in routes.go. Before v0.1.18, I probably would have just killed the separate runner process which would leave the api server intact, but now that it is integrated, that isn't really an option any more.

Originally created by @nperez on GitHub (Jan 21, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/2121 I have a use case where multiple processes (stable diffusion, whsiper, ollama, etc) are competing for limited GPU resources and I need to share the GPU. Unfortunately, there doesn't appear to be a way to manage the session lifetime of loaded models in ollama. It would be cool to have the ability via model options to control the session lifetime (ie. unload after each request) or have a new endpoint to unconditionally unload whatever model is loaded. Without this feature, I need to manage (kill, then restart) the ollama process or wait the five minutes that is the current `defaultSessionDuration` in routes.go. Before v0.1.18, I probably would have just killed the separate runner process which would leave the api server intact, but now that it is integrated, that isn't really an option any more.
GiteaMirror added the feature request label 2026-04-12 10:59:14 -05:00
GiteaMirror commented 2026-04-12 10:59:16 -05:00
Author
Owner
Copy Link

@pdevine commented on GitHub (Jan 27, 2024):

You will be able to use the new keep_alive parameter which was just checked in in #2146 . You can set it to 0 and it will automatically unload the model after inference is completed.

<!-- gh-comment-id:1912861147 --> @pdevine commented on GitHub (Jan 27, 2024): You will be able to use the new `keep_alive` parameter which was just checked in in #2146 . You can set it to `0` and it will automatically unload the model after inference is completed.
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#1212
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?