MEMENTO MORImementomoridev.com

Docs · Guide II

Ollama: local and cloud

One tool runs both sides of the lab. Local and cloud models use the same commands; only the model name changes.

1. Sign in once

Cloud models need an Ollama account. Sign in from a terminal; the Pro plan's cloud models are then available by name.

terminal
$ ollama signin

2. Run a model

Same command for both. Cloud model names end in :cloud, and they don't need to be downloaded first.

terminal
# local — the daily driver
$ ollama run qwen3.8:27b

# cloud — hard coding
$ ollama run glm-5.3:cloud

Not sure which model to run? See Which model for which job.

3. Manage local models

terminal
# download a model
$ ollama pull qwen3.8:27b

# list downloaded models
$ ollama ls

# show models loaded right now
$ ollama ps

# unload a model from memory
$ ollama stop qwen3.8:27b

# delete a downloaded model
$ ollama rm model-name

4. Where model files live

On Windows, Ollama stores models in C:\Users\%username%\.ollama\models by default. This lab keeps them at D:\Ollama\models instead, set with the OLLAMA_MODELS environment variable:

  1. Quit Ollama from its icon in the taskbar.
  2. Open Settings and search for environment variables.
  3. Click Edit environment variables for your account.
  4. Add a variable named OLLAMA_MODELS with the value D:\Ollama\models.
  5. Click OK to save.
  6. Start Ollama again from the Start menu.

5. Keys stay out of code

Calling Ollama's cloud API directly takes an API key. Keys are never committed to code. Anything secret is entered by Tony as an environment variable.

Source: docs.ollama.com