Downloaded models remain on disk even when no model is running. Ollama's installed model list shows their names and tags in the server's local model store.

The CLI and API expose the same inventory through different formats. Each full model name includes a tag, which distinguishes sizes or variants that share a model name.

An empty list means the server is reachable but has no installed models. A connection error instead means the client cannot reach the server, so it says nothing about which models are stored there.

Steps to list Ollama models:

  1. List the models installed on the Ollama server.
    $ ollama list
    NAME                    ID              SIZE      MODIFIED
    granite3.2-vision:2b    3be41a661804    2.4 GB    2 hours ago
    all-minilm:latest       1b226e2802db    45 MB     2 hours ago
    qwen3:0.6b              7df6b6e09427    522 MB    2 hours ago
    qwen2.5:0.5b            a8b0c5157701    397 MB    2 hours ago

    The NAME column contains the complete model name and tag, such as qwen2.5:0.5b. ID is a shortened model digest, SIZE describes the stored model, and MODIFIED changes with time. A store with no installed models returns only the column headings. If a required name is absent, downloading it is a separate task.
    Related: How to pull an Ollama model

    A connection error means the Ollama server is unavailable.
    Related: How to start the Ollama server

  2. Confirm the installed names through the same server's API.
    $ curl -fsS http://localhost:11434/api/tags

    The response's models array contains each model's name, full digest and size in bytes. The names match the CLI's NAME column, and each digest starts with its CLI ID. For qwen2.5:0.5b above, that prefix is a8b0c5157701. An empty store returns {“models”:[]}. If the CLI connects to another server through OLLAMA_HOST, the API URL must address that same server. These inventory requests leave the installed models in place.

Share this guide