Keep a reproducible inventory of local Ollama models
Track installed and loaded models separately so clients use the intended build and you know what is occupying GPU memory.
In brief
- Use `ollama ls` to see locally installed models and `ollama ps` to see running models.
- Keep a separate note of which applications need each model and which should remain available after a restart.
What to check
Use ollama ls to see locally installed models and ollama ps to see running models. A model can be present on disk without occupying VRAM. Use a named tag in the application’s configuration and record the model identifier during acceptance testing. Maintain a small approved inventory on systems shared by multiple applications.
Put it to work
Keep a separate note of which applications need each model and which should remain available after a restart. Test a new tag under a distinct name before switching the client. When reclaiming memory, stopping a model and deleting its weights have different consequences. Confirm that a model is no longer needed before removing it from disk, especially on an isolated host.
Separate the two inventories
The installed-model endpoint returns digests and quantization details. The running-model endpoint reports active models, allocated context and VRAM size. Save both views with the server version after a rollout. A model can remain installed when its runner has unloaded, so audit disk capacity and active memory as distinct resources.
Track what actually changed
Record a mapping from application name to requested tag and observed digest. Compare the mapping before and after pulling or importing models. Re-run a representative prompt when the digest changes, even if the tag remains the same. Keep a record of the prior working version before deleting weights or changing the active tag. This is particularly valuable when the server has limited network access and a rollback download may be slow.
Sources and verification
Based on primary documentation reviewed on 3 November 2026. Practical test procedures are editorial guidance. Performance depends on the model, hardware and configuration.
- CLI Reference Ollama · Primary document
- List models Ollama · Primary document
- List running models Ollama · Primary document
Prepared with AI assistance from the linked sources. This is not a claim of independent field reporting or human review. How we publish · Report a correction
Keep up with homelab
Explore the rest of the local Ollama guide series.
RSS requires an email newsletter opt-in.
