### Install and Run Example Script

Source: https://lemonade-server.ai/docs/dev/omni-router

Install the OpenAI Python client and run the example script to demonstrate the agentic loop with image generation and text-to-speech tools.

```bash
pip install openai
python examples/lemonade_tools.py "Generate an image of a sunset"
python examples/lemonade_tools.py "Say hello world out loud"

```

--------------------------------

### Setup Build Environment (Linux/macOS)

Source: https://lemonade-server.ai/docs/dev/getting-started

Run this script to set up the build environment on Linux and macOS. It installs dependencies using native package managers and creates the build directory.

```bash
./setup.sh

```

--------------------------------

### Start Lemon Zest from Source (Linux/Windows)

Source: https://lemonade-server.ai/docs/integrations/lemon-zest

Start the Lemon Zest application after cloning and installing dependencies. This command runs the application using Node.js.

```bash
npm start

```

--------------------------------

### Install FLM Backend

Source: https://lemonade-server.ai/docs/guide/cli

Installs the FLM backend, potentially for a specific recipe. This command is an example of installing a backend that might not be automatically detected or commonly used.

```bash
lemonade backends install flm:npu
```

--------------------------------

### Ensure Correct Installation with Error Checking

Source: https://lemonade-server.ai/docs/dev/self-hosted-runners

Combine a package installation command with error checking to verify successful installation. This example ensures `pip install -e .` completes without errors.

```powershell
pip install -e .
if ($LASTEXITCODE -ne 0) { exit $LASTEXITCODE }
```

--------------------------------

### Install a backend

Source: https://lemonade-server.ai/docs/guide/cli

Installs a specified backend for the recipe. Ensure the backend name and version are correct.

```bash
lemonade backends install llamacpp:vulkan
```

--------------------------------

### Install Backend (Packaging Time)

Source: https://lemonade-server.ai/docs/embeddable/backends

Installs a specified backend for a given recipe and device. This is useful for bundling backends into your application's installer.

```APIDOC
## lemonade backends install BACKEND:DEVICE

### Description
Installs a specified backend for a given recipe and device. This command is used to bundle backends into your application's installer.

### Usage
```bash
./lemonade backends install llamacpp:vulkan --force
```

### Parameters
* `BACKEND:DEVICE`: The backend to install (e.g., `llamacpp:vulkan`).
* `--force`: Ignores compatibility checks, useful for packaging on a VM.
```

--------------------------------

### Install Lemonade Server via Snap

Source: https://lemonade-server.ai/docs/guide/install/ubuntu

Install the Lemonade server using the snap package manager. This is a convenient way to get the latest version in a sandboxed environment.

```bash
sudo snap install lemonade-server

```

--------------------------------

### Install Backend (Runtime)

Source: https://lemonade-server.ai/docs/embeddable/backends

Installs a specified backend for a given recipe. This is used when the backend choice depends on the user's hardware.

```APIDOC
## POST /v1/install

### Description
Installs a specified backend for a given recipe. This is used when the backend choice depends on the user's hardware.

### Method
POST

### Endpoint
/v1/install

### Parameters
#### Request Body
- **recipe** (string) - Required - The recipe for the backend (e.g., "llamacpp").
- **backend** (string) - Required - The specific backend to install (e.g., "rocm", "vulkan").
- **stream** (boolean) - Optional - Whether to stream the installation process.

### Request Example
```json
{
  "recipe": "llamacpp",
  "backend": "rocm",
  "stream": false
}
```

### Response
#### Success Response (200)
(No specific fields mentioned in the source, typically indicates successful initiation of installation)

#### Response Example
(No example provided in source)
```

--------------------------------

### Install Open WebUI

Source: https://lemonade-server.ai/docs/integrations/open-webui

Install Open WebUI using pip in a dedicated Python environment. Docker installation is also available.

```bash
pip install open-webui
```

--------------------------------

### Install Backend, Bypassing Hardware Filtering

Source: https://lemonade-server.ai/docs/guide/cli

Installs a backend for a recipe even if it's normally filtered due to hardware compatibility checks. Use the `--force` flag to attempt installation regardless of system capabilities.

```bash
lemonade backends install llamacpp:rocm --force
```

--------------------------------

### Install Backend

Source: https://lemonade-server.ai/docs/api/lemonade

Installs a specified backend for a given recipe. Supports streaming progress updates.

```APIDOC
## POST /v1/install

### Description
Installs a backend for a specific recipe. If the `stream` parameter is set to `true`, the endpoint will return Server-Sent Events with installation progress.

### Method
POST

### Endpoint
/v1/install

### Parameters
#### Query Parameters
- **recipe** (string) - Required - Recipe name (for example, `llamacpp`, `flm`, `whispercpp`, `sd-cpp`, `ryzenai-llm`)
- **backend** (string) - Required - Backend name within the recipe (for example, `vulkan`, `rocm`, `cpu`, `default`)
- **stream** (boolean) - Optional - If `true`, returns Server-Sent Events with progress. Defaults to `false`.
- **force** (boolean) - Optional - If `true`, bypasses hardware filtering for `unsupported` backends and attempts installation anyway. Defaults to `false`.

### Request Example
```json
{
  "recipe": "llamacpp",
  "backend": "vulkan",
  "stream": false
}
```

### Response
#### Success Response (200)
- **status** (string) - Indicates the success of the operation.
- **recipe** (string) - The name of the installed recipe.
- **backend** (string) - The name of the installed backend.

#### Response Example
```json
{
  "status": "success",
  "recipe": "llamacpp",
  "backend": "vulkan"
}
```
```

--------------------------------

### Build Windows Installer (WiX/MSI)

Source: https://lemonade-server.ai/docs/dev/getting-started

Use this command to build the WiX/MSI installer for Windows. Ensure WiX Toolset 5.0.2 is installed and the C++ build is completed.

```bash
cmake --build build --config Release --target wix_installers

```

--------------------------------

### Build and Install Lemon Zest Snap Locally

Source: https://lemonade-server.ai/docs/integrations/lemon-zest

Build the Lemon Zest snap package locally on Linux. This involves installing snapcraft, building the snap, and then installing the generated .snap file.

```bash
sudo snap install snapcraft --classic
snapcraft
sudo snap install lemon-zest_*.snap --dangerous

```

--------------------------------

### Setup Build Environment (Windows)

Source: https://lemonade-server.ai/docs/dev/getting-started

Execute this PowerShell script to prepare the build environment on Windows.

```powershell
./setup.ps1

```

--------------------------------

### Install Backend with cURL

Source: https://lemonade-server.ai/docs/api/lemonade

Use this cURL command to install a backend for a specific recipe. Ensure the 'recipe', 'backend', and 'stream' parameters are correctly set.

```bash
curl -X POST http://localhost:13305/v1/install \
  -H "Content-Type: application/json" \
  -d '{
    "recipe": "llamacpp",
    "backend": "vulkan",
    "stream": false
  }'

```

--------------------------------

### POST /v1/install

Source: https://lemonade-server.ai/docs/api/lemonade

Install or update a backend.

```APIDOC
## POST /v1/install

### Description
Install or update a backend.

### Method
POST

### Endpoint
/v1/install
```

--------------------------------

### Install llama.cpp backend on Windows

Source: https://lemonade-server.ai/docs/embeddable/backends

Use this command to install the llama.cpp backend with Vulkan support on Windows. Ensure `lemond` is running and downloading to the specified directory.

```batch
REM Start lemond to download backends to ./bin/
lemond.exe ./

REM Download llama.cpp with the Vulkan backend to ./bin/llamacpp/vulkan
lemonade.exe backends install llamacpp:vulkan
```

--------------------------------

### Install Vulkan backend on Windows

Source: https://lemonade-server.ai/docs/embeddable/backends

Use this command to install the Vulkan backend for llama.cpp on Windows as a fallback if ROCm is not supported. The `stream: false` option indicates that the installation should not be streamed.

```bash
curl -X POST http://localhost:8000/v1/install ^
  -H "Content-Type: application/json" ^
  -d "{\"recipe\": \"llamacpp\", \"backend\": \"vulkan\", \"stream\": false}"
```

--------------------------------

### Install Lemon Zest via Snap

Source: https://lemonade-server.ai/docs/integrations/lemon-zest

Install Lemon Zest using the Snap package manager on Linux. This command installs the application directly from the Snap Store.

```bash
sudo snap install lemon-zest

```

--------------------------------

### Example config.json Structure

Source: https://lemonade-server.ai/docs/guide/configuration

This is a comprehensive example of the `config.json` file, illustrating various settings for server behavior, model loading, and backend configurations for different AI models.

```json
{
  "config_version": 1,
  "port": 13305,
  "host": "localhost",
  "log_level": "info",
  "global_timeout": 300,
  "max_loaded_models": 1,
  "no_broadcast": false,
  "extra_models_dir": "",
  "models_dir": "auto",
  "ctx_size": 4096,
  "offline": false,
  "no_fetch_executables": false,
  "disable_model_filtering": false,
  "enable_dgpu_gtt": false,
  "rocm_channel": "preview",
  "llamacpp": {
    "backend": "auto",
    "args": "",
    "prefer_system": false,
    "rocm_bin": "builtin",
    "vulkan_bin": "builtin",
    "cpu_bin": "builtin"
  },
  "whispercpp": {
    "backend": "auto",
    "args": "",
    "cpu_bin": "builtin",
    "npu_bin": "builtin"
  },
  "sdcpp": {
    "backend": "auto",
    "args": "",
    "steps": 20,
    "cfg_scale": 7.0,
    "width": 512,
    "height": 512,
    "cpu_bin": "builtin",
    "rocm_bin": "builtin",
    "vulkan_bin": "builtin"
  },
  "flm": {
    "args": "",
  },
  "ryzenai": {
    "server_bin": "builtin"
  },
  "kokoro": {
    "cpu_bin": "builtin"
  }
}

```

--------------------------------

### Install Linux .deb Package

Source: https://lemonade-server.ai/docs/dev/getting-started

Install the generated .deb package using 'apt'. Replace <VERSION> with the actual version number of the package.

```bash
# Replace <VERSION> with the actual version (e.g., 9.0.0)
sudo apt install ./lemonade-server_<VERSION>_amd64.deb

```

--------------------------------

### Install Vulkan backend on Linux

Source: https://lemonade-server.ai/docs/embeddable/backends

Use this command to install the Vulkan backend for llama.cpp on Linux as a fallback if ROCm is not supported. The `stream: false` option indicates that the installation should not be streamed.

```bash
curl -X POST http://localhost:8000/v1/install \
  -H "Content-Type: application/json" \
  -d '{
    "recipe": "llamacpp",
    "backend": "vulkan",
    "stream": false
  }'
```

--------------------------------

### Install Linux .rpm Package

Source: https://lemonade-server.ai/docs/dev/getting-started

Install the generated .rpm package using 'dnf'. Replace <VERSION> with the actual version number.

```bash
# Replace <VERSION> with the actual version (e.g., 9.0.0)
sudo dnf install ./lemonade-server-<VERSION>.x86_64.rpm

```

--------------------------------

### Get runtime configuration snapshot

Source: https://lemonade-server.ai/docs/embeddable/runtime

Demonstrates how to retrieve the full runtime configuration of the `lemond` instance using a GET request to the `/internal/config` endpoint.

```bash
curl http://localhost:8000/internal/config

```

--------------------------------

### Install a Model

Source: https://lemonade-server.ai/docs/api/lemonade

Installs a model that is already registered with the Lemonade Server. If the model is not registered, it can also register and install a model from Hugging Face.

```APIDOC
## POST /v1/pull

### Description
Installs a registered model or registers and installs a model from Hugging Face.

### Method
POST

### Endpoint
/v1/pull

### Parameters
#### Common Parameters
- **stream** (boolean) - No - If `true`, returns Server-Sent Events (SSE) with download progress. Defaults to `false`.

#### Request Body
- **model_name** (string) - Yes - Lemonade Server model name to install or namespaced model name to register and install (e.g., `user.Phi-4-Mini-GGUF`).
- **checkpoint** (string) - Yes - HuggingFace checkpoint to install (required when registering a new model).
- **recipe** (string) - Yes - Lemonade API recipe to load the model with (required when registering a new model).
- **reasoning** (boolean) - No - Whether the model is a reasoning model (default: false).
- **vision** (boolean) - No - Whether the model has vision capabilities (default: false).
- **embedding** (boolean) - No - Whether the model is an embedding model (default: false).
- **reranking** (boolean) - No - Whether the model is a reranking model (default: false).
- **mmproj** (string) - No - Multimodal Projector (mmproj) file to use for vision models.

### Request Example
```json
{
  "model_name": "Qwen2.5-0.5B-Instruct-CPU"
}
```

### Request Example (Register and Install)
```json
{
  "model_name": "user.Phi-4-Mini-GGUF",
  "checkpoint": "unsloth/Phi-4-mini-instruct-GGUF:Q4_K_M",
  "recipe": "llamacpp"
}
```

### Response
#### Success Response (200)
- **status** (string) - Indicates success or error.
- **message** (string) - Confirmation message, e.g., "Installed model: [model_name]".

#### Response Example
```json
{
  "status": "success",
  "message": "Installed model: Qwen2.5-0.5B-Instruct-CPU"
}
```

### Streaming Response (stream=true)
When `stream=true`, the endpoint returns Server-Sent Events with real-time download progress.

**Event Types:**
- `progress`: Sent during download with current file and byte progress.
- `complete`: Sent when all files are downloaded successfully.
- `error`: Sent if download fails, with `error` field containing the message.
```

--------------------------------

### Clone and Install Lemon Zest from Source (Linux/Windows)

Source: https://lemonade-server.ai/docs/integrations/lemon-zest

Clone the Lemon Zest repository from GitHub and install its Node.js dependencies. This is the method for running the application from its source code.

```bash
git clone https://github.com/phqen1x/lemon-zest.git
cd lemon-zest
npm install

```

--------------------------------

### Install llama.cpp backend on Linux

Source: https://lemonade-server.ai/docs/embeddable/backends

Use this command to install the llama.cpp backend with Vulkan support on Linux. The `--force` option bypasses system compatibility checks.

```bash
# Start lemond to download backends to ./bin/
./lemond ./

# Download llama.cpp with the Vulkan backend to ./bin/llamacpp/vulkan
./lemonade backends install llamacpp:vulkan --force
```

--------------------------------

### Install lemonade-server on Arch Linux

Source: https://lemonade-server.ai/docs/guide/install/arch

Use this command to install the lemonade-server package from the Arch User Repository (AUR) using yay. Ensure you have yay installed and configured.

```bash
yay -S lemonade-server

```

--------------------------------

### Register and Install a New Model from Hugging Face

Source: https://lemonade-server.ai/docs/api/lemonade

Register a new model from Hugging Face and install it. The `model_name` must be namespaced with `user.` to avoid conflicts. Specify the `checkpoint` from Hugging Face and the `recipe` for loading the model.

```bash
curl -X POST http://localhost:13305/v1/pull \
  -H "Content-Type: application/json" \
  -d '{ \
    "model_name": "user.Phi-4-Mini-GGUF", \
    "checkpoint": "unsloth/Phi-4-mini-instruct-GGUF:Q4_K_M", \
    "recipe": "llamacpp" \
  }'
```

--------------------------------

### Install ROCm backend on Windows

Source: https://lemonade-server.ai/docs/embeddable/backends

Use this command to install the ROCm backend for llama.cpp on Windows if ROCm is supported by your system. The `stream: false` option indicates that the installation should not be streamed.

```bash
curl -X POST http://localhost:8000/v1/install \
  -H "Content-Type: application/json" \
  -d "{\"recipe\": \"llamacpp\", \"backend\": \"rocm\", \"stream\": false}"
```

--------------------------------

### Launch Open WebUI Server

Source: https://lemonade-server.ai/docs/integrations/open-webui

Run this command to start the Open WebUI HTTP server. Access it via http://localhost:8080/ in your browser.

```bash
open-webui serve
```

--------------------------------

### Run Lemonade Server

Source: https://lemonade-server.ai/docs/integrations/lemon-zest

Start the Lemonade Server with the Flux-2-Klein-4B model. The server will download the model automatically on first launch and listen on http://localhost:13305. Ensure the server is running before starting Lemon Zest.

```bash
lemonade-server run Flux-2-Klein-4B

```

--------------------------------

### Install vLLM ROCm Backend

Source: https://lemonade-server.ai/docs/guide/configuration/vllm

Installs the vLLM ROCm backend using the lemonade CLI. Ensure you have the necessary prerequisites met.

```bash
lemonade backends install vllm:rocm

```

--------------------------------

### Install a Registered Model with Lemonade Server

Source: https://lemonade-server.ai/docs/api/lemonade

Use this endpoint to install a model that is already registered with the Lemonade Server. Ensure the `model_name` corresponds to an existing registered model.

```bash
curl -X POST http://localhost:13305/v1/pull \
  -H "Content-Type: application/json" \
  -d '{ \
    "model_name": "Qwen2.5-0.5B-Instruct-CPU" \
  }'
```

--------------------------------

### Install Lemonade Desktop App via Snap

Source: https://lemonade-server.ai/docs/guide/install/ubuntu

Install the fully sandboxed Lemonade desktop application using snap. This provides a self-contained desktop experience.

```bash
sudo snap install lemonade

```

--------------------------------

### Install Webpack and Tauri CLI Dependencies

Source: https://lemonade-server.ai/docs/dev/app

Installs the necessary Node.js packages for the webpack build and Tauri CLI. Run this command within the `src/app` directory.

```bash
npm ci
```

--------------------------------

### Get system info on Windows

Source: https://lemonade-server.ai/docs/embeddable/backends

Call this endpoint to retrieve system information, including supported backend recipes and their states, on Windows.

```bash
curl http://localhost:8000/v1/system-info
```

--------------------------------

### Install Lemonade Desktop App via Apt

Source: https://lemonade-server.ai/docs/guide/install/ubuntu

Install the Lemonade desktop application using apt, which includes the web UI. This package is useful for users who want a desktop integration.

```bash
sudo apt install lemonade-desktop

```

--------------------------------

### Start Docker Compose Services

Source: https://lemonade-server.ai/docs/guide/install/docker

Execute this command in the directory containing your `docker-compose.yml` file to pull the image and start the Lemonade server in detached mode.

```bash
docker-compose up -d

```

--------------------------------

### Docker Run with Default Configuration

Source: https://lemonade-server.ai/docs/guide/install/docker

Use this command to start the Lemonade server with default settings and persistent volumes for caching.

```bash
docker run -d \
  --name lemonade-server \
  -p 13305:13305 \
  -v lemonade-cache:/root/.cache/huggingface \
  -v lemonade-llama:/opt/lemonade/llama \
  -v lemonade-recipe:/root/.cache/lemonade \
  ghcr.io/lemonade-sdk/lemonade-server:latest

```

--------------------------------

### Install Lemonade Server via PPA

Source: https://lemonade-server.ai/docs/guide/install/ubuntu

Add the stable PPA for Lemonade and install the server package. This method is suitable for users who prefer managing packages through apt.

```bash
sudo add-apt-repository ppa:lemonade-team/stable
sudo apt install lemonade-server

```

--------------------------------

### Install Lemonade Desktop on Arch Linux

Source: https://lemonade-server.ai/docs/guide/install/arch

Install the Lemonade Desktop package, which provides the graphical user interface, from the AUR using yay. This package makes the web app available at http://localhost:13305.

```bash
yay -S lemonade-desktop

```

--------------------------------

### Install a Vision Language Model (VLM)

Source: https://lemonade-server.ai/docs/integrations/open-webui

Install a VLM through the Lemonade Model Manager by navigating to the Model Management tab and clicking the '+' button next to a model with a 'VISION' label.

```html
http://localhost:13305
```

--------------------------------

### Verify Linux Executables

Source: https://lemonade-server.ai/docs/dev/getting-started

After installation, the executables 'lemonade' and 'lemond' should be available in the system's PATH.

```bash
lemonade --help
lemond --help

```

--------------------------------

### Example curl request with Authorization header

Source: https://lemonade-server.ai/docs/embeddable/runtime

Shows an example of using `curl` to make a GET request to the `/v1/health` endpoint, including the required `Authorization` header.

```bash
curl http://localhost:8000/v1/health \
  -H "Authorization: Bearer KEY"

```

--------------------------------

### List Available Recipes and Backends

Source: https://lemonade-server.ai/docs/guide/cli

Lists all available recipes and their corresponding backends that can be installed. This command helps in understanding system compatibility and available options.

```bash
lemonade backends
```

--------------------------------

### Pull a registered model

Source: https://lemonade-server.ai/docs/guide/cli

Downloads and installs a model from the Lemonade Server registry. This is the simplest way to get a model.

```bash
lemonade pull Qwen3-0.6B-GGUF
```

--------------------------------

### Configure Custom Vulkan Binary Path (Linux)

Source: https://lemonade-server.ai/docs/embeddable/backends

Use these commands in bash to start `lemond` and then set the path to your custom Vulkan `llama-server` binary.

```bash
# Start lemond to update configuration
./lemond ./

# Set the llama-server vulkan binary path
./lemonade config set llamacpp.vulkan_bin /path/to/bins

```

--------------------------------

### Get System Info

Source: https://lemonade-server.ai/docs/embeddable/backends

Retrieves information about the system's supported backends and their states. This is crucial for determining which backends can be installed.

```APIDOC
## GET /v1/system-info

### Description
Retrieves information about the system's supported backends and their states. This is crucial for determining which backends can be installed.

### Method
GET

### Endpoint
/v1/system-info

### Response
#### Success Response (200)
- **recipes** (object) - Information about available recipes and their backends.
  - **llamacpp** (object)
    - **backends** (object)
      - **rocm** (object)
        - **devices** (array) - List of supported ROCm devices.
        - **state** (string) - Installation state of the ROCm backend ('installable', 'installed', etc.).
      - **vulkan** (object) - Information about the Vulkan backend.

### Response Example
```json
{
  "recipes": {
    "llamacpp": {
      "backends": {
        "rocm": {
          "devices": ["amd_igpu"],
          "state": "installable"
        }
      }
    }
  }
}
```
```

--------------------------------

### Configure Custom Vulkan Binary Path (Windows)

Source: https://lemonade-server.ai/docs/embeddable/backends

Use these commands in Windows cmd.exe to start `lemond` and then set the path to your custom Vulkan `llama-server` binary.

```batch
REM Start lemond to update configuration
lemond.exe ./

REM Set the llama-server vulkan binary path
lemonade.exe config set llamacpp.vulkan_bin C:\path\to\bins

```

--------------------------------

### Install Claude Code using npm

Source: https://lemonade-server.ai/docs/integrations/claude-code

Installs Claude Code globally using npm. Requires Node.js and npm to be installed.

```bash
npm install -g @anthropic-ai/claude-code

```

--------------------------------

### Install or Update Backend

Source: https://lemonade-server.ai/docs/api/lemonade

Installs or updates a backend for a specific recipe. If the backend is already installed but outdated, this endpoint updates it to the configured version.

```APIDOC
## POST /v1/install

### Description
Install or update a backend for a specific recipe/backend pair. If the backend is already installed but outdated, this endpoint updates it to the configured version.

### Method
POST

### Endpoint
/v1/install
```

--------------------------------

### Install vLLM ROCm Backend via API

Source: https://lemonade-server.ai/docs/guide/configuration/vllm

Installs the vLLM ROCm backend using the Lemonade API. This is an alternative to the CLI installation.

```bash
curl -X POST http://localhost:13305/api/v1/install \
  -H 'Content-Type: application/json' \
  -d '{"recipe": "vllm", "backend": "rocm"}'

```

--------------------------------

### Launch an Agent with Default Model Settings

Source: https://lemonade-server.ai/docs/guide/cli

Launches a specified agent with default model settings. Use this for basic agent startup.

```bash
lemonade launch claude --model Qwen3.5-0.8B-GGUF
```

--------------------------------

### Load Model with RyzenAI LLM Options

Source: https://lemonade-server.ai/docs/guide/cli

Load a model using the 'ryzenai-llm' recipe. The primary option available is customizing the context size.

```bash
lemonade load MODEL_NAME --ctx-size SIZE
```

--------------------------------

### Install ROCm backend on Linux

Source: https://lemonade-server.ai/docs/embeddable/backends

Use this command to install the ROCm backend for llama.cpp on Linux if ROCm is supported by your system. The `stream: false` option indicates that the installation should not be streamed.

```bash
curl -X POST http://localhost:8000/v1/install \
  -H "Content-Type: application/json" \
  -d '{
    "recipe": "llamacpp",
    "backend": "rocm",
    "stream": false
  }'
```

--------------------------------

### Install Claude Code using curl

Source: https://lemonade-server.ai/docs/integrations/claude-code

Installs Claude Code using a curl command. Ensure you have bash available.

```bash
curl -fsSL https://claude.ai/install.sh | bash

```

--------------------------------

### Launch Agent with Prompt-Driven Recipe Import

Source: https://lemonade-server.ai/docs/guide/cli

Launches an agent and enables prompt-driven recipe import using remote directory and recipe file flags. This is for importing recipes at prompt time.

```bash
lemonade launch claude --directory coding-agents --recipe-file Qwen3.5-35B-A3B-NoThinking.json
```

--------------------------------

### Build macOS from Source

Source: https://lemonade-server.ai/docs/dev/getting-started

Instructions for building Lemonade Server from source on macOS for M-Series/arm64. This includes installing Xcode command line tools, configuring CMake, and building the release version.

```bash
# Install Xcode command line tools
xcode-select --install

# Navigate to the C++ source directory
cd src/cpp

# Create and enter build directory
mkdir build
cd build

# Configure with CMake
cmake ..

# Build with all cores
cmake --build . --config Release -j

```

--------------------------------

### Display Server Models List (Linux/macOS)

Source: https://lemonade-server.ai/docs/embeddable/models

Print the current list of suggested models available to the server on Linux or macOS. This list is defined in `server_models.json`.

```bash
./lemond ./ 
./lemonade list
```

--------------------------------

### Install Lemonade RPM on Fedora

Source: https://lemonade-server.ai/docs/guide/install/fedora

Use this command to install the downloaded Lemonade RPM package on Fedora. Ensure you have the correct path to the downloaded file.

```bash
sudo dnf install ./DOWNLOADED_FILE
```

--------------------------------

### Load Model with SD.cpp Options

Source: https://lemonade-server.ai/docs/guide/cli

Load a model for image generation using the 'sd-cpp' recipe. Configure the backend, custom arguments, inference steps, guidance scale, and image dimensions.

```bash
lemonade load MODEL_NAME --sdcpp BACKEND --sdcpp-args ARGS --steps N --cfg-scale SCALE --width PX --height PX
```

--------------------------------

### Build Embeddable Server with Web App (Windows)

Source: https://lemonade-server.ai/docs/embeddable/building

Builds the embeddable server and CLI for Windows, including web app assets. The `web-app` target must be built before `embeddable`.

```bash
cmake --preset windows -DBUILD_WEB_APP=ON
cmake --build --preset windows --target web-app embeddable
```

--------------------------------

### Load Model with FLM Options

Source: https://lemonade-server.ai/docs/guide/cli

Load a model using the 'flm' recipe. Adjust context size and provide custom arguments for the 'flm serve' command.

```bash
lemonade load MODEL_NAME --ctx-size SIZE --flm-args ARGS
```

--------------------------------

### Load Model with Llama.cpp Options

Source: https://lemonade-server.ai/docs/guide/cli

Load a model using the 'llamacpp' recipe. Customize context size and specify the backend or custom arguments for llama-server.

```bash
lemonade load MODEL_NAME --ctx-size SIZE --llamacpp BACKEND --llamacpp-args ARGS
```

--------------------------------

### Docker Run with CPU Backend

Source: https://lemonade-server.ai/docs/guide/install/docker

Start the Lemonade server explicitly using the CPU backend for llama.cpp, ensuring compatibility and performance on systems without GPU support.

```bash
docker run -d \
  --name lemonade-server \
  -p 13305:13305 \
  -v lemonade-cache:/root/.cache/huggingface \
  -v lemonade-llama:/opt/lemonade/llama \
  -v lemonade-recipe:/root/.cache/lemonade \
  -e LEMONADE_LLAMACPP=cpu \
  ghcr.io/lemonade-sdk/lemonade-server:latest

```

--------------------------------

### Start Lemonade Server

Source: https://lemonade-server.ai/docs/dev/getting-started

Execute the lemond server with default options or specify a custom port and host address. Available options include port, host, version, and help.

```bash
# Start server with default options
./lemond

```

```bash
# Start server with custom port
./lemond --port 8080

```

--------------------------------

### Build Tauri App on Windows (Visual Studio 2022)

Source: https://lemonade-server.ai/docs/dev/getting-started

Execute this command to build the Tauri desktop app using Visual Studio 2022 on Windows.

```bash
cmake --build --preset windows --target tauri-app

```

--------------------------------

### Build and Notarize macOS Package

Source: https://lemonade-server.ai/docs/dev/getting-started

Navigate to the build directory and use CMake to build a release configuration and then notarize the generated macOS package.

```bash
cd src/cpp/build
cmake --build . --config Release --target package-macos
cmake --build . --config Release --target notarize_package
```

--------------------------------

### Run Model with Custom Context and Open Web App

Source: https://lemonade-server.ai/docs/guide/cli

Loads a model with a custom context size and then opens the Lemonade Web App. This combines model configuration with immediate access to the UI.

```bash
lemonade run Qwen3-0.6B-GGUF --ctx-size 8192
```

--------------------------------

### Launch an Agent with a Specific Llama.cpp Backend

Source: https://lemonade-server.ai/docs/guide/cli

Launches an agent using a specific Llama.cpp backend. Useful for optimizing performance on different hardware.

```bash
lemonade launch codex --model Qwen3.5-0.8B-GGUF --llamacpp vulkan
```

--------------------------------

### GET /v1/system-info

Source: https://lemonade-server.ai/docs/api/lemonade

System information and device enumeration.

```APIDOC
## GET /v1/system-info

### Description
System information and device enumeration.

### Method
GET

### Endpoint
/v1/system-info
```

--------------------------------

### GET /v1/stats

Source: https://lemonade-server.ai/docs/api/lemonade

Performance statistics from the last request.

```APIDOC
## GET /v1/stats

### Description
Performance statistics from the last request.

### Method
GET

### Endpoint
/v1/stats
```

--------------------------------

### Use a custom llama.cpp build

Source: https://lemonade-server.ai/docs/guide/configuration

Use your own locally built llama.cpp executable. Lemonade will use the specified binary directly and will not perform any downloads.

```bash
lemonade config set llamacpp.vulkan_bin=/home/me/llama.cpp/build/bin
```

--------------------------------

### Display Server Models List (Windows)

Source: https://lemonade-server.ai/docs/embeddable/models

Print the current list of suggested models available to the server on Windows. This list is defined in `server_models.json`.

```batch
lemond.exe ./ 
lemonade.exe list
```

--------------------------------

### GET /v1/health

Source: https://lemonade-server.ai/docs/api/lemonade

Check server status, such as models loaded.

```APIDOC
## GET /v1/health

### Description
Check server status, such as models loaded.

### Method
GET

### Endpoint
/v1/health
```