Skip to main content

Quickstart Guide

Get your first local model running in less than a minute. GoingMerry makes executing state-of-the-art models on consumer boundaries completely effortless.

1. Start the Server​

Before running any model operations, ensure the GoingMerry daemon is serving requests. If you installed the application bundle, this daemon runs automatically in the background.

If running manually via the CLI, start the daemon:

merry serve

2. Execute a Model​

To download and start talking to a model in a single step, use the merry run command followed by the model identifier.

For example, to execute Google's Gemma 4 (or Llama 4, DeepSeek, etc.):

merry run gemma4

Behind the scenes, GoingMerry:

  1. Checks if the model weights are cached locally.
  2. If not, pulls the sharded tensor layers from the registry.
  3. Allocates graphics memory and compiles execution kernels.
  4. Opens an interactive chat prompt.
$ merry run gemma4
pulling manifest 
pulling 8f8c7e9b23b8... 100% ▕████████████████████████████████████████████████████████▏ 4.7 GB
pulling 43070624a1b0... 100% ▕████████████████████████████████████████████████████████▏  15 KB
pulling c3a1b024b898... 100% ▕████████████████████████████████████████████████████████▏  11 KB
success 

>>> Write a quick python bubble sort function
Here is a simple implementation of Bubble Sort in Python:

def bubble_sort(arr):
    n = len(arr)
    for i in range(n):
        for j in range(0, n-i-1):
            if arr[j] > arr[j+1]:
                arr[j], arr[j+1] = arr[j+1], arr[j]
    return arr

>>>

3. Interactive Console Shortcuts​

While inside the interactive CLI prompt, you can use several slash (/) commands to control the session:

CommandAction
/bye or /exitTerminate the prompt and release model memory.
/help or /?Print help and list available prompt shortcuts.
/set parameter <key> <val>Dynamically update inference parameters (e.g. /set parameter temperature 0.7).
/show infoInspect the active model parameters, architecture, and licensing tags.

4. Run a One-Off Prompt​

If you want to pipe data into GoingMerry or run a non-interactive shell query, you can pass the prompt text directly in the command arguments:

merry run gemma4 "Explain quantum computing in one sentence"

You can also pipe text files directly into models:

cat code_snippet.py | merry run gemma4 "Review this code for memory leaks"