Expose "Processor" on PS() Call

[mattyMattMatt]

I am trying to get information on where the model is running in the system. On the CLI this retrieved with ollama ps. Here it displays:

PS C:\Users\m> ollama ps
NAME               ID              SIZE      PROCESSOR    UNTIL
llama3.2:latest    a80c4f17acd5    4.0 GB    100% GPU     2 minutes from now

It looks like in the API call "Processor" is not exposed which is why when running python ollama.ps() what I get is:
model='llama3.2:latest' name='llama3.2:latest' digest='a80c4f17acd55265feec403c7aef86be0c25983ab279d83f3bcd3abbcb5b8b72' expires_at=datetime.datetime(2025, 3, 31, 16, 13, 25, 974861, tzinfo=TzInfo(-04:00)) size=3972362240 size_vram=3972362240 details=ModelDetails(parent_model='', format='gguf', family='llama', families=['llama'], parameter_size='3.2B', quantization_level='Q4_K_M')

I have trawled through documents without finding where this is exposed so I assume it is not? I dont really want to run a subprocess call to the ollama cli but this information is rather useful for testing aspects of integration performance between systems. Any help would be appreciated.

[TommyBoiss]

You can get the percentage its running on the GPU by dividing size_vram by size then timing it by 100,
Example:

gpu_percentage = (size_vram/ size) * 100
cpu_percentage = 100 - gpu_percentage 

[mattyMattMatt]

Thank you that is quite elegant and obvious when thinking about it. I appreciate the concise answer.