Model Management

Manage your local LiteRT-LM models using the CLI.

LiteRT-LM stores imported models in a local registry. Once imported, you can refer to these models by their Model ID in other commands (like run or serve).

List Models

To view all imported models in your local registry:

litert-lm list

This will output a table showing the ID, SIZE, and MODIFIED timestamp for each model.

ID              SIZE            MODIFIED
gemma3-1b       557.3 MB        2026-03-05 17:00:53
gemma4-e2b      2.4 GB          2026-04-02 10:44:33
gemma4-e4b      3.4 GB          2026-04-02 10:45:10

Import a Model

You can import .litertlm models into your local registry from a local path or directly from the Hugging Face Hub.

The basic syntax is: bash litert-lm import SOURCE_PATH [MODEL_ID] If MODEL_ID is omitted, the filename of the source model will be used as the default ID.

Import from a Local Path

To import a model file located on your machine:

litert-lm import gemma-4-E2B-it.litertlm gemma4-e2b

Import from Hugging Face

To download and import a model directly from a Hugging Face repository:

litert-lm import \
  --from-huggingface-repo litert-community/gemma-4-E2B-it-litert-lm \
  gemma-4-E2B-it.litertlm \
  gemma4-e2b

litert-lm import `
  --from-huggingface-repo litert-community/gemma-4-E2B-it-litert-lm `
  gemma-4-E2B-it.litertlm `
  gemma4-e2b

Note: If the repository is private, you can specify your Hugging Face token using the --huggingface-token option.

Run the Imported Model

After importing, you can use the model ID to run the model.

litert-lm run gemma4-e2b

Rename a Model

To rename an existing model in your local registry:

litert-lm rename old-model-id new-model-id

This will safely move the model files to the new identifier. The command will fail if the new ID is already in use.

Delete a Model

To delete a model and free up disk space:

litert-lm delete my-model

This permanently removes the model files from your local registry.