Ollama CPU or GPU? Share what ollama ps reports

If you are investigating model placement, Ollama's official FAQ documents ollama ps for inspecting currently loaded models. Its processor column indicates CPU, GPU, or split placement.

For a useful help request, collect the output while the relevant model is loaded and include:

• Ollama version and the exact model identifier.
• Operating system, GPU, available GPU memory, and system RAM.
• The context setting and other concurrent model requests.
• A short, sanitized prompt and the behavior you are investigating.
• Relevant server-log messages from the same attempt.

If you share a screenshot, check it for private paths and other sensitive details. Avoid inferring a cause from placement alone; explain the symptom and compare observations from a controlled test.

What does your setup report, and which part of the behavior would you like help understanding?