[GH-ISSUE #8478] Display Minimum System Requirements #31219

Closed
opened 2026-04-22 11:27:41 -05:00 by GiteaMirror · 6 comments
Owner

Originally created by @Siddhesh-Agarwal on GitHub (Jan 18, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/8478

It would be great to have the minimum system requirements like disk space, and RAM for each model. This could be done both on the website and in the CLI like a small notice.

The minimum RAM required by the model is:

minimum_RAM = num_of_parameters * bytes_per_parameter

I see that most models use the Q4_K_M or Q4_0 quantization. So, bytes_per_parameter = 0.5. I may be oversimplifying the complexity of this problem.

Originally created by @Siddhesh-Agarwal on GitHub (Jan 18, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/8478 It would be great to have the minimum system requirements like disk space, and RAM for each model. This could be done both on the website and in the CLI like a small notice. The minimum RAM required by the model is: minimum_RAM = num_of_parameters * bytes_per_parameter I see that most models use the `Q4_K_M` or `Q4_0` quantization. So, bytes_per_parameter = $0.5$. I may be oversimplifying the complexity of this problem.
GiteaMirror added the feature request label 2026-04-22 11:27:41 -05:00
Author
Owner

@Siddhesh-Agarwal commented on GitHub (Jan 18, 2025):

The reason I say I may be oversimplifying the problem is because I am not talking about the overhead of Ollama itself (or is that not required?)

<!-- gh-comment-id:2599508692 --> @Siddhesh-Agarwal commented on GitHub (Jan 18, 2025): The reason I say I may be oversimplifying the problem is because I am not talking about the overhead of Ollama itself (or is that not required?)
Author
Owner

@rick-github commented on GitHub (Jan 18, 2025):

Image

<!-- gh-comment-id:2599508875 --> @rick-github commented on GitHub (Jan 18, 2025): ![Image](https://github.com/user-attachments/assets/5faac306-6a52-4496-80de-40985818e0f4)
Author
Owner

@Siddhesh-Agarwal commented on GitHub (Jan 18, 2025):

Sorry, I wasn't able to figure it out. Thanks for your help!

<!-- gh-comment-id:2599511864 --> @Siddhesh-Agarwal commented on GitHub (Jan 18, 2025): Sorry, I wasn't able to figure it out. Thanks for your help!
Author
Owner

@adarshmadrecha commented on GitHub (Jan 18, 2025):

@rick-github thanks for pointing out.
In this case, there should be a tooltip to inform users that this "43GB" is the RAM requirement and not the file size.

<!-- gh-comment-id:2599863324 --> @adarshmadrecha commented on GitHub (Jan 18, 2025): @rick-github thanks for pointing out. In this case, there should be a tooltip to inform users that this "43GB" is the RAM requirement and not the file size.
Author
Owner

@rick-github commented on GitHub (Jan 18, 2025):

It's both.

<!-- gh-comment-id:2599930056 --> @rick-github commented on GitHub (Jan 18, 2025): It's both.
Author
Owner

@Siddhesh-Agarwal commented on GitHub (Jan 19, 2025):

@adarshmadrecha the model in its entirety is loaded in the RAM. So the model size is the space it takes in the RAM

<!-- gh-comment-id:2600722869 --> @Siddhesh-Agarwal commented on GitHub (Jan 19, 2025): @adarshmadrecha the model in its entirety is loaded in the RAM. So the model size is the space it takes in the RAM
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#31219