github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 08:30:05 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #7041] Variable OLLAMA_MAX_LOADED_MODELS is being ignored #50979

New Issue
Closed
opened 2026-04-28 17:44:11 -05:00 by GiteaMirror · 1 comment
GiteaMirror commented 2026-04-28 17:44:11 -05:00
Owner
Copy Link

Originally created by @hdnh2006 on GitHub (Sep 30, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/7041

Originally assigned to: @dhiltgen on GitHub.

What is the issue?

Hello!

Thanks for this fantastic framework you have created.

I'm using Llama for deploying my private LLMs, but I'm experiencing an issue where it seems like the environment variable is being ignored. I'm not sure if I'm setting it incorrectly, and I'd appreciate any help in troubleshooting the issue.

I am setting Environment="OLLAMA_MAX_LOADED_MODELS=2" (or 3 or 4, no matter the number) in the config file /etc/systemd/system/ollama.service and after restarting the service with sudo systemctl daemon-reload and sudo systemctl restart ollama just one model is being loaded.

I have an RTX 4060 with 16GB VRAM but I still have 64 GB of RAM available to run the second model on CPU, but it looks like this doesn't work.

Am I missing something here? is it possible to load model on both CPU and GPU at the same time?

Thanks in advance.

OS

Linux

GPU

Nvidia

CPU

AMD

Ollama version

ollama version is 0.3.12

Originally created by @hdnh2006 on GitHub (Sep 30, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/7041 Originally assigned to: @dhiltgen on GitHub. ### What is the issue? Hello! Thanks for this fantastic framework you have created. I'm using Llama for deploying my private LLMs, but I'm experiencing an issue where it seems like the environment variable is being ignored. I'm not sure if I'm setting it incorrectly, and I'd appreciate any help in troubleshooting the issue. I am setting `Environment="OLLAMA_MAX_LOADED_MODELS=2"` (or 3 or 4, no matter the number) in the config file `/etc/systemd/system/ollama.service` and after restarting the service with `sudo systemctl daemon-reload` and `sudo systemctl restart ollama` just one model is being loaded. I have an RTX 4060 with 16GB VRAM but I still have 64 GB of RAM available to run the second model on CPU, but it looks like this doesn't work. Am I missing something here? is it possible to load model on both CPU and GPU at the same time? Thanks in advance. ### OS Linux ### GPU Nvidia ### CPU AMD ### Ollama version ollama version is 0.3.12
GiteaMirror added the feature request label 2026-04-28 17:44:11 -05:00
GiteaMirror commented 2026-04-28 17:44:13 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Sep 30, 2024):

Ollama currently prioritizes loading models in the GPU. It will evict models from the GPU to load a new one if both models won't fit in the GPU. If models will fit, it won't load more than OLLAMA_MAX_LOADED_MODELS in GPU. To force loading models into RAM, you need to set num_gpu to 0. See https://github.com/ollama/ollama/issues/6950#issuecomment-2373663650 for details (and the other comments in the ticket for other thoughts on model management).

<!-- gh-comment-id:2382722267 --> @rick-github commented on GitHub (Sep 30, 2024): Ollama currently prioritizes loading models in the GPU. It will evict models from the GPU to load a new one if both models won't fit in the GPU. If models will fit, it won't load more than `OLLAMA_MAX_LOADED_MODELS` in GPU. To force loading models into RAM, you need to set `num_gpu` to 0. See https://github.com/ollama/ollama/issues/6950#issuecomment-2373663650 for details (and the other comments in the ticket for other thoughts on model management).
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#50979
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?