System Monitor
The System Monitor view provides real-time hardware metrics for your machine, helping you understand resource usage while running local models. Access it from the activity icon in the sidebar orCtrl+Shift+M.
Overview
The top of the monitor view shows a summary dashboard with the most important metrics at a glance:
- CPU usage percentage and core utilization graph
- RAM usage (used / total) with a progress bar
- GPU usage, VRAM usage, and GPU temperature (if available)
- Disk read/write throughput
- Number of running models and their combined memory footprint
CPU
The CPU section shows per-core utilization as a bar chart, overall CPU percentage as a time-series graph (last 60 seconds), and the current clock speed. If the model is running on CPU, this section highlights the cores handling inference.
Memory (RAM)
Displays total system RAM, currently used, and available. A real-time graph tracks memory usage over time. When a model is loaded, the estimated model memory footprint is shown separately.
GPU
If a compatible GPU is detected, the GPU section shows:
Running Models
The bottom section lists currently loaded models with their memory usage, context length, and inference engine. Each model entry shows tokens per second throughput during active generation.
Use the stop button next to a model to unload it from memory and free VRAM. Unloading a model does not delete it or its conversation history.
Hardware Compatibility
OMM includes a hardware compatibility checker that runs on first launch and can be re-run from the monitor view. It checks:
- CPU architecture and instruction set support (AVX2, NEON)
- Available RAM vs model requirements
- GPU detection and driver compatibility
- Disk space for model storage
- Recommended models based on your hardware