My LLM user guide

This guide covers installation, entering text, copying results, and everyday use. The screenshots show the English interface on macOS. On Windows, use the system tray instead of the menu bar.

For an introduction to the app, see My LLM. To look up individual settings, see the settings reference.

If you are new to My LLM, follow this page from the beginning. If you already use it, jump to everyday use, creating a task, hotkeys and window controls, or troubleshooting.

1. Install the app and prepare a provider

Install My LLM

Download the file for your operating system and processor from GitHub Releases.

System	File to choose	Installation
macOS 13 or later, Apple Silicon	My-LLM-vX.X.X-darwin-arm64.dmg	Open the DMG and drag My LLM.app to Applications
macOS 13 or later, Intel	My-LLM-vX.X.X-darwin-x86_64.dmg	Same as above
Windows 11, x86_64	My-LLM-vX.X.X-windows-x86_64.zip	Extract the ZIP and run myllm.exe. Keep settings.exe in the same folder

vX.X.X is the version number. The macOS distribution is signed and notarized. The Windows ZIP is unsigned.

When started normally, My LLM stays in the menu bar or system tray. If no window appears immediately, choose Open Window from the icon's menu.

Prepare a model

Choose one of the following options.

Option	Preparation
Local Ollama model	Start Ollama and download a model. Set its name and endpoint in My LLM
Ollama cloud model	Sign in to Ollama and specify a cloud model name. Use it through your locally running Ollama instance
OpenAI or Anthropic API	Obtain an API key for the service and the name of the model you want to use

For Ollama, follow the official quickstart. My LLM's initial configuration uses llama3.2 for regular tasks and translategemma:12b for translation. To use local models, install Ollama and obtain the models separately from My LLM.

If the Ollama command is available, you can download the local models as follows. To use a model you already have, set the Model field in Settings to its name.

ollama pull llama3.2
ollama pull translategemma:12b

Memory requirements and speed vary by model. You can start with a model for regular tasks and set up translation later.

Use Ollama cloud models

To use a cloud model through locally running Ollama, sign in to Ollama and make sure the model is available to your account. In My LLM, keep Provider set to ollama and Base URL set to http://localhost:11434, then change Model to the cloud model name. To sign in, run:

ollama signin

Use a cloud model name listed by Ollama for its local app and CLI. For example, the official documentation lists gemma4:cloud. Check the cloud model documentation for available names. You do not need to download the cloud model's weights to your computer.

Ollama handles cloud authentication for this connection. Even though My LLM connects to localhost, your input and instructions are sent to the selected cloud model.

2. Run your first task

1. Choose Settings… from the menu. In Providers, set the endpoint and Default model you want to use. For a cloud API, configure its API key as well.
2. In Tasks, set the task's Provider to that endpoint. If Model is blank, it uses the provider's Default model. Start with a sample task such as polish or summarize.
3. Close Settings, then choose Open Window from the menu.
4. Paste a short passage into Input.
5. Choose the task you prepared at the bottom left and click Run.
6. When the result appears in Output, click Copy and paste it back into your original app.

Providers holds endpoints and their default models; Tasks holds the instructions to apply to your text. Translation has separate provider and model settings in Translation. You can check that a regular task works before setting up translation.

Figure 1: An example summary of photography workshop notes. Input is at the top and Output below. The task and model are at the bottom left, with Run and Copy at the bottom right.

Control	Purpose
Input	Type or paste the text to process. Editing is disabled while a task is running
Output	Shows the result as it arrives. You can select the displayed text
Task selector	Chooses the instructions to use. Selecting a task alone does not run it
Provider and model label	Shows the provider and model for the current task
Run	Processes the text in Input with the selected task
Copy	Copies the result in Output to the clipboard

The result appears progressively. A waiting indicator may appear before the first response. Results depend on the model and instructions, so review them before use.

3. Everyday use

Paste text before running

Open Window opens the window with empty Input and Output fields. It does not read the clipboard or run a task. Use it when you want to type text yourself or check what you have pasted before running it.

Choose the task initially selected by Open Window using Open Window task in General.

Copy text and call a task

1. Copy text in your usual app.
2. Choose a task from the My LLM menu, or press its hotkey.
3. The window opens and automatically processes the clipboard text.

Figure 2: An example menu. Open Window opens an empty window, while individual tasks capture the input and run it. Task names and hotkeys depend on your settings.

Menu item	Action
Open Window	Opens empty Input and Output fields
Individual tasks / Translate	Captures input and runs the selected task
Reload Config	Reloads the configuration file and updates menus and hotkeys
Open Config Folder	Opens the folder containing the configuration file
Settings…	Opens the settings window
Quit My LLM	Quits the app, including its menu bar or tray process

Capture selected text directly

Enable Send ⌘/Control+C to copy the selection. in General to send a copy command to the frontmost app when you call a task. My LLM then captures the selected text. When this option is off, it uses the text already on the clipboard.

On macOS, this copy action requires Accessibility permission for My LLM. Go to System Settings → Privacy & Security → Accessibility. See Apple's instructions.

If permission is missing or the frontmost app cannot copy the selection, My LLM may use the existing clipboard contents instead. Starting by copying manually makes it easier to check the input.

Copy results automatically

Enable automatic copying in General to place successfully completed results on the clipboard. You can override this for individual tasks, for example to copy polished text automatically but read explanations in the window.

The priority is the individual task or translation setting, then General, then enabled as the fallback. The Copy button works independently of automatic copying.

4. Open and save settings

Choose Settings… from the menu. The settings window has four panes: General, Providers, Tasks, and Translation.

On macOS, valid changes are saved automatically. On Windows, click OK or Apply to save. After saving, close Settings so that My LLM reloads the changes. If you edit the configuration file directly, choose Reload Config from the menu.

For the meaning and possible values of each setting, see the settings reference.

Create a task

1. In Settings… → Tasks, click + and set the task ID and Name. The ID is an identifier such as workshop_summary; Name is the label shown in the menu.
2. Write what you want the model to do and how to format the result in Instruction. For example, save instructions for a bullet-point summary or your preferred writing style.
3. Choose Provider and, if needed, Model. Leave Model blank to use that provider's Default model.
4. Set Hotkey and automatic copying if needed, save, and close Settings.
5. Choose the task in Open Window and try a short passage. Adjust Instruction after reviewing the result.

See Tasks in the settings reference for an example instruction and the meaning of each field. Configure translation instructions in Translation, rather than adding a task in Tasks.

Find a setting

What you want to configure	Details
Appearance, interface language, automatic copying, and default task	General
Endpoints, models, and API keys	Providers
Task names, instructions, and individual hotkeys	Tasks
Translation engine and language routing	Translation
Edit the configuration file directly	File location and TOML example
Call the GUI from a launcher	Launch the GUI from a command

5. Hotkeys and window controls

Enter the key combination you want to record in a hotkey field. Choose combinations that do not conflict with the operating system or other apps.

Action	macOS	Windows
Run the text in Input	⌘R	Ctrl+R
Copy Output	⌘C	Ctrl+C
Close the window	⌘W or Esc	Ctrl+W or Esc
Quit the app	⌘Q or Quit My LLM in the menu	Ctrl+Q or Quit My LLM in the tray menu

When you select text in Input or Output and copy it, copying the selection takes priority. To copy the entire result, use the Copy button.

In normal resident mode, closing the window does not quit the app. Hiding the window while a task is running does not cancel the task. To exit the app, choose Quit My LLM.

Troubleshooting

Problem	What to check
No window appears after launch	In resident mode, choose Open Window from the menu
A task is missing	Check its entry and Instruction in Tasks. Enable translation in Translation
An endpoint or model error appears	Check Provider and Model in Providers and Tasks. For Ollama, also check that it is running and the model is downloaded
Cloud authentication fails	Check the API key and selected provider. If using Environment, also check the app's launch environment
The input differs from the text you copied	Check whether selection capture is enabled. Try copying manually before calling the task
Selected text is not captured	Check macOS Accessibility permission and whether the frontmost app can copy the text
A hotkey does not respond	Check the recorded combination and conflicts with the OS or other apps. Also try running the task from the menu
Run is disabled	A task is running, or both the input and task match the last submission. Enter new text or choose another task
Results are not copied automatically	Check whether the task overrides General. Results are not copied automatically when an error occurs
Settings changes do not take effect	Close Settings after saving. For direct file edits, choose Reload Config
Settings… does not open	Extract the download again and check that the settings app is included. On Windows, keep both EXE files in the same folder

The model names and task names shown in the screenshots are examples. The layout and basic operations have been checked on macOS. The Windows instructions are based on the code and bundled README; no screenshots were taken on a Windows machine for this guide.