### Install Vertex AI SDK

Source: https://github.com/googleapis/python-aiplatform/blob/main/README.rst

Install the Vertex AI SDK for Python using pip or uv.

```console
pip install google-cloud-aiplatform
```

```console
uv pip install google-cloud-aiplatform
```

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

### Install Dependencies

Source: https://github.com/googleapis/python-aiplatform/blob/main/scripts/readme-gen/templates/install_deps.tmpl.rst

Install all the required Python packages listed in the requirements.txt file using pip.

```bash
$ pip install -r requirements.txt
```

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

### Install Vertex AI SDK with Prediction Support (Windows)

Source: https://github.com/googleapis/python-aiplatform/blob/main/google/cloud/aiplatform/prediction/README.md

Installs the virtualenv tool, creates a virtual environment, activates it, and installs the Vertex AI SDK with prediction support. Replace `<your-env>` with your desired environment folder name.

```shell
pip install virtualenv
virtualenv <your-env>
<your-env>\Scripts\activate
<your-env>\Scripts\pip.exe install "google-cloud-aiplatform[prediction]>=1.16.0"
```

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

### Install Vertex AI SDK

Source: https://github.com/googleapis/python-aiplatform/blob/main/gemini_docs/README.md

Install the google-cloud-aiplatform Python package with the specified version. Use the --upgrade --user flags for installation.

```shell
pip3 install --upgrade --user "google-cloud-aiplatform>=1.114.0"
```

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

### Install Vertex AI SDK with Prediction Support (Mac/Linux)

Source: https://github.com/googleapis/python-aiplatform/blob/main/google/cloud/aiplatform/prediction/README.md

Installs the virtualenv tool, creates a virtual environment, activates it, and installs the Vertex AI SDK with prediction support. Replace `<your-env>` with your desired environment folder name.

```shell
pip install virtualenv
virtualenv <your-env>
source <your-env>/bin/activate
<your-env>/bin/pip install "google-cloud-aiplatform[prediction]>=1.16.0"
```

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

### Install google-cloud-aiplatform SDK

Source: https://github.com/googleapis/python-aiplatform/blob/main/agentplatform/model_garden/README.md

Install or upgrade the google-cloud-aiplatform Python package to the latest version.

```shell
pip3 install --upgrade --user "google-cloud-aiplatform>=1.84"
```

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

### Get Help for Python Sample

Source: https://github.com/googleapis/python-aiplatform/blob/main/scripts/readme-gen/templates/README.tmpl.rst

This command displays the help message for a Python sample script, providing information on its usage and available arguments.

```bash
{{get_help(sample.file)|indent}}
```

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

### Fast Tryout Deployment

Source: https://github.com/googleapis/python-aiplatform/blob/main/agentplatform/model_garden/README.md

Enable an experimental fast-deploy path for popular models, allowing for interactive experimentation without a full production setup.

```python
endpoint = model.deploy(fast_tryout_enabled=True)
```

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

### Generative Model with System Instructions

Source: https://github.com/googleapis/python-aiplatform/blob/main/gemini_docs/README.md

Configures a generative model with system instructions to guide its behavior and response style. This example sets the model to talk like a pirate and avoid rude words.

```python
from vertexai.generative_models import GenerativeModel
model = GenerativeModel(
    "gemini-1.0-pro",
    system_instruction=[
        "Talk like a pirate.",
        "Don\'t use rude words.",
    ],
)
print(model.generate_content("Why is sky blue?"))
```

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

### Install Vertex AI SDK

Source: https://github.com/googleapis/python-aiplatform/blob/main/vertexai/generative_models/README.md

Install or upgrade the google-cloud-aiplatform Python package to version 1.38 or higher.

```shell
pip3 install --upgrade --user "google-cloud-aiplatform>=1.38"
```

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

### Install Vertex AI SDK

Source: https://github.com/googleapis/python-aiplatform/blob/main/vertexai/model_garden/README.md

Install the google-cloud-aiplatform Python package to use the SDK. Ensure you have a recent version.

```shell
pip3 install --upgrade --user "google-cloud-aiplatform>=1.84"
```

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

### Install PyAudio on Windows

Source: https://github.com/googleapis/python-aiplatform/blob/main/scripts/readme-gen/templates/install_portaudio.tmpl.rst

On Windows, PortAudio is typically installed automatically when PyAudio is installed, often without explicit steps.

```bash
pip install pyaudio
```

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

### Install virtualenv and google-cloud-aiplatform on Windows

Source: https://github.com/googleapis/python-aiplatform/blob/main/docs/README.rst

Installs the virtualenv tool and the google-cloud-aiplatform library in a virtual environment on Windows systems.

```bash
pip install virtualenv
virtualenv <your-env>
<your-env>\Scripts\activate
<your-env>\Scripts\pip.exe install google-cloud-aiplatform
```

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

### Install virtualenv and google-cloud-aiplatform on Mac/Linux

Source: https://github.com/googleapis/python-aiplatform/blob/main/docs/README.rst

Installs the virtualenv tool and the google-cloud-aiplatform library in a virtual environment on Mac or Linux systems.

```bash
pip install virtualenv
virtualenv <your-env>
source <your-env>/bin/activate
<your-env>/bin/pip install google-cloud-aiplatform
```

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

### Configure System Instructions for Model

Source: https://github.com/googleapis/python-aiplatform/blob/main/vertexai/generative_models/README.md

Customize model behavior by providing system instructions during model initialization. This example sets instructions to 'Talk like a pirate.' and 'Don't use rude words.'

```python
from vertexai.generative_models import GenerativeModel
model = GenerativeModel(
    "gemini-1.0-pro",
    system_instruction=[
        "Talk like a pirate.",
        "Don't use rude words.",
    ],
)
print(model.generate_content("Why is sky blue?"))
```

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

### Quick Start: Default Deployment

Source: https://github.com/googleapis/python-aiplatform/blob/main/agentplatform/model_garden/README.md

Deploys a model using the default configuration. This is suitable for fast prototyping and initial evaluation of model outputs.

```APIDOC
## Quick Start: Default Deployment

This is the simplest way to deploy a model. If you provide just a model name, the SDK will use the default deployment configuration.

### Code Example
```python
from agentplatform import model_garden

model = model_garden.OpenModel("google/paligemma@paligemma-224-float32")
endpoint = model.deploy()
```

### Use Case
Fast prototyping, first-time users evaluating model outputs.
```

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

### Quick Start: Default Model Deployment

Source: https://github.com/googleapis/python-aiplatform/blob/main/agentplatform/model_garden/README.md

Deploy a model using the default configuration by providing only the model name. Ideal for fast prototyping and initial evaluations.

```python
from agentplatform import model_garden

model = model_garden.OpenModel("google/paligemma@paligemma-224-float32")
endpoint = model.deploy()
```

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

### Install PortAudio on Mac OS X

Source: https://github.com/googleapis/python-aiplatform/blob/main/scripts/readme-gen/templates/install_portaudio.tmpl.rst

Use Homebrew to install PortAudio on macOS. This is a prerequisite for PyAudio.

```bash
brew install portaudio
```

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

### Troubleshoot PyAudio Installation on Mac OS X

Source: https://github.com/googleapis/python-aiplatform/blob/main/scripts/readme-gen/templates/install_portaudio.tmpl.rst

If pip install encounters errors related to 'portaudio.h', use these flags to specify include and library paths.

```bash
pip install --global-option='build_ext'     --global-option='-I/usr/local/include'     --global-option='-L/usr/local/lib'     pyaudio
```

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

### Install PortAudio on Debian/Ubuntu Linux

Source: https://github.com/googleapis/python-aiplatform/blob/main/scripts/readme-gen/templates/install_portaudio.tmpl.rst

Install the PortAudio development package and Python development headers on Debian-based Linux distributions.

```bash
apt-get install portaudio19-dev python-all-dev
```

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

### Install Agent Engine Packages

Source: https://github.com/googleapis/python-aiplatform/blob/main/gemini_docs/README.md

Install the necessary packages for Agent Engine and ADK. Ensure you are using version 1.111 or higher.

```shell
pip3 install --upgrade --user "google-cloud-aiplatform[agent_engines,adk]>=1.111"
```

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

### Install Python Development Headers

Source: https://github.com/googleapis/python-aiplatform/blob/main/CONTRIBUTING.rst

If you encounter errors related to 'Python.h' not being found, install the python-dev package on Debian/Ubuntu systems.

```bash
sudo apt-get install python-dev
```

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

### Function Calling with Tools

Source: https://github.com/googleapis/python-aiplatform/blob/main/vertexai/generative_models/README.md

Implement function calling by defining tools with function declarations. This example defines a 'get_current_weather' function and uses it in a chat.

```python
# First, create tools that the model is can use to answer your questions.
# Describe a function by specifying it's schema (JsonSchema format)
get_current_weather_func = generative_models.FunctionDeclaration(
    name="get_current_weather",
    description="Get the current weather in a given location",
    parameters={
        "type": "object",
        "properties": {
            "location": {
                "type": "string",
                "description": "The city and state, e.g. San Francisco, CA"
            },
            "unit": {
                "type": "string",
                "enum": [
                    "celsius",
                    "fahrenheit",
                ]
            }
        },
        "required": [
            "location"
        ]
    },
)
# Tool is a collection of related functions
weather_tool = generative_models.Tool(
    function_declarations=[get_current_weather_func],
)

# Use tools in chat:
model = GenerativeModel(
    "gemini-pro",
    # You can specify tools when creating a model to avoid having to send them with every request.
    tools=[weather_tool],
)
chat = model.start_chat()
# Send a message to the model. The model will respond with a function call.
print(chat.send_message("What is the weather like in Boston?"))
# Then send a function response to the model. The model will use it to answer.
print(chat.send_message(
    Part.from_function_response(
        name="get_current_weather",
        response={
            "content": {"weather": "super nice"},
        }
    ),
))
```

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

### Get a Model

Source: https://github.com/googleapis/python-aiplatform/blob/main/README.rst

Retrieve a model resource using its resource name.

```Python
model = aiplatform.Model('/projects/my-project/locations/us-central1/models/{MODEL_ID}')
```

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

### Generate Content with System Instructions

Source: https://github.com/googleapis/python-aiplatform/blob/main/gemini_docs/README.md

Generates content with specific system instructions to guide the model's behavior.

```APIDOC
## Generate Content with System Instructions
### Description
Generates content with specific system instructions to guide the model's behavior.

### Method
```python
from vertexai.generative_models import GenerativeModel
model = GenerativeModel(
    "gemini-1.0-pro",
    system_instruction=[
        "Talk like a pirate.",
        "Don't use rude words.",
    ],
)
print(model.generate_content("Why is sky blue?"))
```
```

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

### Get Specific Skill Revision

Source: https://github.com/googleapis/python-aiplatform/blob/main/README.rst

Retrieves a specific skill revision by its resource name. This example gets the first revision from the list.

```Python
# Get a specific skill revision by its resource name
if revisions_response.skill_revisions:
    target_revision_name = revisions_response.skill_revisions[0].name
    revision = client.skills.revisions.get(name=target_revision_name)
```

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

### Create a Client

Source: https://github.com/googleapis/python-aiplatform/blob/main/README.rst

Demonstrates how to initialize a Gen AI client for Vertex AI.

```APIDOC
## Create a Client

### Description
Initializes a Gen AI client for use with Vertex AI, specifying the project and location.

### Method
`genai.Client()`

### Parameters
- **vertexai** (bool) - Required - Set to `True` to use Vertex AI.
- **project** (str) - Required - The Google Cloud project ID.
- **location** (str) - Required - The Google Cloud location.

### Request Example
```python
genai_client = genai.Client(vertexai=True, project="your-project", location="your-location")
```
```

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

### Run Python Sample

Source: https://github.com/googleapis/python-aiplatform/blob/main/scripts/readme-gen/templates/README.tmpl.rst

Use this command to execute a Python sample script. Ensure you have the necessary API enabled and role permissions.

```bash
$ python {{sample.file}}
```

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

### Get Predictions

Source: https://github.com/googleapis/python-aiplatform/blob/main/docs/README.rst

Gets predictions from a deployed model on an endpoint.

```APIDOC
## Get Predictions

### Description
Gets predictions from a deployed model on an endpoint.

### Method
`Endpoint.predict()`

### Parameters
*   **instances** (list) - Required - A list of instances to get predictions for.

### Request Example
```python
endpoint.predict(instances=[[6.7, 3.1, 4.7, 1.5], [4.6, 3.1, 1.5, 0.2]])
```
```

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

### Grounding Responses with Google Search

Source: https://github.com/googleapis/python-aiplatform/blob/main/vertexai/generative_models/README.md

Configure a GenerativeModel to use Google Search for retrieving information to ground its responses. This example shows how to set up the tool and print the response text and grounding metadata.

```python
from vertexai import generative_models
from vertexai.generative_models import GenerativeModel, Tool

model=GenerativeModel(
    "gemini-pro",
    tools=[Tool.from_google_search_retrieval(
        google_search_retrieval=generative_models.grounding.GoogleSearchRetrieval(
            dynamic_retrieval_config=generative_models.grounding.DynamicRetrievalConfig(
                mode=generative_models.grounding.DynamicRetrievalConfig.Mode.MODE_DYNAMIC))) 
    ],
)

response = model.generate_content("Who won Euro 2024")

print(response.text)

# Checking grounding metadata. It contains grounding supports and the follow-up search entry widget.
if response.candidates:
    print(response.candidates[0].grounding_metadata)
```

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

### Get Predictions from Endpoint

Source: https://github.com/googleapis/python-aiplatform/blob/main/README.rst

Sends instances to an endpoint to get predictions.

```APIDOC
## Get Predictions from Endpoint

### Description
This section describes how to send prediction requests to a deployed endpoint, providing instances and receiving predictions.

### Method
`endpoint.predict`

### Parameters
- `instances` (list) - A list of instances for which to get predictions.

### Request Example
```python
endpoint.predict(instances=[[6.7, 3.1, 4.7, 1.5], [4.6, 3.1, 1.5, 0.2]])
```
```

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

### Initialize Vertex AI SDK

Source: https://github.com/googleapis/python-aiplatform/blob/main/docs/README.rst

Initialize the Vertex AI SDK with common configurations like project ID, location, and staging bucket. Environment defaults are used if parameters are not provided.

```Python
aiplatform.init(
    # your Google Cloud Project ID or number
    # environment default used is not set
    project='my-project',

    # the Vertex AI region you will use
    # defaults to us-central1
    location='us-central1',

    # Google Cloud Storage bucket in same region as location
    # used to stage artifacts
    staging_bucket='gs://my_staging_bucket',

    # custom google.auth.credentials.Credentials
    # environment default credentials used if not set
    credentials=my_credentials,

    # customer managed encryption key resource name
    # will be applied to all Vertex AI resources if set
    encryption_spec_key_name=my_encryption_key_name,

    # the name of the experiment to use to track
    # logged metrics and parameters
    experiment='my-experiment',

    # description of the experiment above
    experiment_description='my experiment description'
)
```

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

### Build HTML Documentation

Source: https://github.com/googleapis/python-aiplatform/blob/main/CONTRIBUTING.rst

Build the project's HTML documentation using nox. This command should be run after making documentation changes.

```bash
nox -s docs
```

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

### Initialize Vertex AI SDK

Source: https://github.com/googleapis/python-aiplatform/blob/main/README.rst

Initialize the SDK with common configurations like project ID, region, and GCS bucket. Ensure credentials and encryption spec are set if needed.

```Python
aiplatform.init(
    # your Google Cloud Project ID or number
    # environment default used is not set
    project='my-project',

    # the Vertex AI region you will use
    # defaults to us-central1
    location='us-central1',

    # Google Cloud Storage bucket in same region as location
    # used to stage artifacts
    staging_bucket='gs://my_staging_bucket',

    # custom google.auth.credentials.Credentials
    # environment default credentials used if not set
    credentials=my_credentials,

    # customer managed encryption key resource name
    # will be applied to all Vertex AI resources if set
    encryption_spec_key_name=my_encryption_key_name,

    # the name of the experiment to use to track
    # logged metrics and parameters
    experiment='my-experiment',

    # description of the experiment above
    experiment_description='my experiment description'
)
```

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

### Run All Sample Tests

Source: https://github.com/googleapis/python-aiplatform/blob/main/CONTRIBUTING.rst

Navigate to the samples/snippets directory and run all tests for a specific Python version using nox.

```bash
cd samples/snippets
nox -s py-3.9
```

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

### Select a Verified Deployment: By Container and Hardware

Source: https://github.com/googleapis/python-aiplatform/blob/main/agentplatform/model_garden/README.md

Deploys a model by specifying both a serving container image URI and a hardware configuration.

```APIDOC
## Select a Verified Deployment: By Container and Hardware

Specify both a container image and a hardware configuration from the list of verified deployment configurations.

### Code Example
```python
endpoint = model.deploy(
    serving_container_image_uri="us-docker.pkg.dev/vertex-ai/vertex-vision-model-garden-dockers/pytorch-vllm-serve:20250430_0916_RC00_maas",
    machine_type="a3-highgpu-1g",
    accelerator_type="NVIDIA_H100_80GB",
    accelerator_count=1,
)
```

**Use case:** Production configuration, performance tuning, scaling.
```

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

### Fast Tryout Deployment

Source: https://github.com/googleapis/python-aiplatform/blob/main/agentplatform/model_garden/README.md

Enables an experimental fast-deploy path for popular models by setting `fast_tryout_enabled=True`.

```APIDOC
## Fast Tryout Deployment

Enable experimental fast-deploy path for popular models.

### Code Example
```python
endpoint = model.deploy(fast_tryout_enabled=True)
```

**Use case:** Interactive experimentation without full production setup.
```

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

### Deploy Model with Container and Hardware

Source: https://github.com/googleapis/python-aiplatform/blob/main/vertexai/model_garden/README.md

Deploy a model by specifying both a verified container image URI and hardware configuration. This is recommended for production environments and performance optimization.

```python
endpoint = model.deploy(
    serving_container_image_uri="us-docker.pkg.dev/vertex-ai/vertex-vision-model-garden-dockers/pytorch-vllm-serve:20250430_0916_RC00_maas",
    machine_type="a3-highgpu-1g",
    accelerator_type="NVIDIA_H100_80GB",
    accelerator_count=1,
)
```

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

### Create and Activate Virtual Environment

Source: https://github.com/googleapis/python-aiplatform/blob/main/scripts/readme-gen/templates/install_deps.tmpl.rst

Create a new virtual environment named 'env' and activate it. Samples are compatible with Python 3.7+.

```bash
$ virtualenv env
$ source env/bin/activate
```

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

### Get Predictions from Endpoint

Source: https://github.com/googleapis/python-aiplatform/blob/main/README.rst

Send instances to an endpoint to get predictions. The input format should match the deployed model's requirements.

```Python
endpoint.predict(instances=[[6.7, 3.1, 4.7, 1.5], [4.6, 3.1, 1.5, 0.2]])
```

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

### Deploy Model with Container and Hardware

Source: https://github.com/googleapis/python-aiplatform/blob/main/agentplatform/model_garden/README.md

Deploy a model by specifying both a serving container image URI and hardware configuration. Suitable for production and performance tuning.

```python
endpoint = model.deploy(
    serving_container_image_uri="us-docker.pkg.dev/vertex-ai/vertex-vision-model-garden-dockers/pytorch-vllm-serve:20250430_0916_RC00_maas",
    machine_type="a3-highgpu-1g",
    accelerator_type="NVIDIA_H100_80GB",
    accelerator_count=1,
)
```

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

### Install Pre-commit Hooks

Source: https://github.com/googleapis/python-aiplatform/blob/main/CONTRIBUTING.rst

Enable pre-commit hooks to automatically enforce linter checks during the commit process. This command installs the hooks into your local repository's .git/hooks directory.

```bash
pre-commit install
pre-commit installed at .git/hooks/pre-commit
```

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

### Get Skill

Source: https://github.com/googleapis/python-aiplatform/blob/main/gemini_docs/README.md

Retrieves an existing skill by its resource name.

```APIDOC
## get

### Description
Retrieves an existing skill by its resource name.

### Method
client.skills.get

### Parameters
#### Query Parameters
- **name** (string) - Required - The resource name of the skill (e.g., 'projects/{project_id_or_number}/locations/{location}/skills/{skill_id}').

### Response
#### Success Response (200)
- **skill_id** (string) - The ID of the skill.
- **name** (string) - The resource name of the skill.
- **display_name** (string) - The display name of the skill.
- **description** (string) - The description of the skill.
```

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

### Get Skill

Source: https://github.com/googleapis/python-aiplatform/blob/main/README.rst

Retrieves an existing skill by its resource name.

```APIDOC
## Get Skill

### Description
Fetches an existing skill from the Skill Registry using its resource name.

### Method
`client.skills.get()`

### Parameters
- **name** (str) - Required - The resource name of the skill to retrieve.

### Request Example
```python
fetched_skill = client.skills.get(name=skill.name)
```
```

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

### List Model Deployment Options

Source: https://github.com/googleapis/python-aiplatform/blob/main/agentplatform/model_garden/README.md

Inspect the available deployment configurations, including compatible machine types and container images, for a specific model.

```python
model = model_garden.OpenModel("google/paligemma@paligemma-224-float32")
deploy_options = model.list_deploy_options()
```

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

### Create Prompt (Dictionary Format)

Source: https://github.com/googleapis/python-aiplatform/blob/main/gemini_docs/README.md

Define a prompt using a dictionary structure, including prompt data, system instructions, variables, and the model name. Then, use `client.prompts.create()` to register it.

```python
prompt = {
    "prompt_data": {
        "contents": [{"parts": [{"text": "Hello, {name}! How are you?"}]}],
        "system_instruction": {"parts": [{"text": "Please answer in a short sentence."}]},
        "variables": [
            {"name": {"text": "Alice"}},
        ],
        "model": "gemini-2.5-flash",
    },
}

prompt_resource = client.prompts.create(
    prompt=prompt,
)
```

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

### Get Prompt

Source: https://github.com/googleapis/python-aiplatform/blob/main/gemini_docs/README.md

Retrieves an existing prompt resource by its ID.

```APIDOC
## get

### Description
Retrieves an existing prompt resource by its ID.

### Method
client.prompts.get

### Parameters
#### Query Parameters
- **prompt_id** (string) - Required - The ID of the prompt to retrieve.

### Response
#### Success Response (200)
- **prompt_id** (string) - The ID of the retrieved prompt.
- **prompt_data** (object) - The data of the prompt.
  - **model** (string) - The model used for the prompt.
```

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

### Get a Model

Source: https://github.com/googleapis/python-aiplatform/blob/main/README.rst

Retrieves a model resource using its resource name.

```APIDOC
## Get a Model

### Description
This section describes how to retrieve a model resource from Vertex AI using its full resource name.

### Method
`aiplatform.Model(model_name)`

### Parameters
- `model_name` (str) - The resource name of the model (e.g., 'projects/my-project/locations/us-central1/models/{MODEL_ID}').

### Request Example
```python
model = aiplatform.Model('/projects/my-project/locations/us-central1/models/{MODEL_ID}')
```
```

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

### Get an Existing Skill

Source: https://github.com/googleapis/python-aiplatform/blob/main/README.rst

Retrieves an existing skill by its resource name.

```Python
fetched_skill = client.skills.get(name=skill.name)
```

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

### Select a Verified Deployment: By Hardware

Source: https://github.com/googleapis/python-aiplatform/blob/main/agentplatform/model_garden/README.md

Deploys a model using a specified hardware configuration, including machine type, accelerator type, and count.

```APIDOC
## Select a Verified Deployment: By Hardware

Specify a hardware configuration from the list of verified deployment configurations.

### Code Example
```python
endpoints = model.deploy(
    machine_type="a3-highgpu-1g",
    accelerator_type="NVIDIA_H100_80GB",
    accelerator_count=1,
)
```
```

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

### Get Model Evaluation

Source: https://github.com/googleapis/python-aiplatform/blob/main/README.rst

Retrieves the model evaluation resource for a given model.

```APIDOC
## Get Model Evaluation

### Description
This section explains how to retrieve a specific model evaluation resource for a given model. It can retrieve the first evaluation by default or a specific one using its ID.

### Method
`model.get_model_evaluation`

### Parameters
- `evaluation_id` (str, optional) - The ID of the specific model evaluation to retrieve.

### Request Example
```python
model = aiplatform.Model('projects/my-project/locations/us-central1/models/{MODEL_ID}')

# returns the first evaluation with no arguments, you can also pass the evaluation ID
evaluation = model.get_model_evaluation()

eval_metrics = evaluation.metrics
```
```

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

### Run a Single Sample Test

Source: https://github.com/googleapis/python-aiplatform/blob/main/CONTRIBUTING.rst

Navigate to the samples/snippets directory and run a specific sample test by its name using nox.

```bash
cd samples/snippets
nox -s py-3.9 -- -k <name of test>
```

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

### Get Skill Revision

Source: https://github.com/googleapis/python-aiplatform/blob/main/gemini_docs/README.md

Retrieves a specific skill revision by its resource name.

```APIDOC
## Get Skill Revision
### Description
Retrieves a specific skill revision by its resource name.

### Method
```python
client.skills.revisions.get(name=target_revision_name)
```

### Parameters
#### Path Parameters
- **name** (str) - Required - The resource name of the skill revision.
```

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

### Get Skill Revision

Source: https://github.com/googleapis/python-aiplatform/blob/main/README.rst

Retrieves a specific revision of a skill by its resource name.

```APIDOC
## Get Skill Revision

### Description
Fetches a specific revision of a skill using its resource name.

### Method
`client.skills.revisions.get()`

### Parameters
- **name** (str) - Required - The resource name of the skill revision to retrieve.

### Request Example
```python
revision = client.skills.revisions.get(name=target_revision_name)
```
```

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

### Deploy Model with Explanation Metadata

Source: https://github.com/googleapis/python-aiplatform/blob/main/README.rst

Use explanation metadata when deploying a model to an endpoint or uploading a model. The metadata is obtained using `builder.get_metadata_protobuf()`.

```Python
explanation_metadata = builder.get_metadata_protobuf()

# To deploy a model to an endpoint with explanation
model.deploy(..., explanation_metadata=explanation_metadata)

# To deploy a model to a created endpoint with explanation
endpoint.deploy(..., explanation_metadata=explanation_metadata)

# To upload a model with explanation
aiplatform.Model.upload(..., explanation_metadata=explanation_metadata)
```

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

### Import Vertex AI SDK

Source: https://github.com/googleapis/python-aiplatform/blob/main/gemini_docs/README.md

Imports the Vertex AI SDK library. This is the first step before initializing the SDK.

```python
import vertexai
```

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

### Get Model Evaluation

Source: https://github.com/googleapis/python-aiplatform/blob/main/docs/README.rst

Retrieves a specific model evaluation resource for a given model.

```APIDOC
## Get Model Evaluation

### Description
Retrieves a specific model evaluation resource for a given model.

### Method
`Model.get_model_evaluation()`

### Parameters
*   **evaluation_id** (str) - Optional - The ID of the model evaluation to retrieve.

### Request Example
```python
model = aiplatform.Model('projects/my-project/locations/us-central1/models/{MODEL_ID}')
evaluation = model.get_model_evaluation()
```
```

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

### Deploy Model on Spot VM

Source: https://github.com/googleapis/python-aiplatform/blob/main/agentplatform/model_garden/README.md

Deploy a model on Spot VMs to reduce costs. This is beneficial for cost-sensitive development and batch workloads.

```python
endpoint = model.deploy(spot=True)
```

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

### Get Model Evaluation

Source: https://github.com/googleapis/python-aiplatform/blob/main/README.rst

Retrieve a specific model evaluation resource. Can fetch the first evaluation or by evaluation ID.

```Python
model = aiplatform.Model('projects/my-project/locations/us-central1/models/{MODEL_ID}')

# returns the first evaluation with no arguments, you can also pass the evaluation ID
evaluation = model.get_model_evaluation()

eval_metrics = evaluation.metrics
```

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

### Optimize Prompt (Data-Driven)

Source: https://github.com/googleapis/python-aiplatform/blob/main/gemini_docs/README.md

For data-driven prompt optimization, use the `optimize` method with `vapo_config`. This requires specifying either a service account or project number and a config path. Logging is recommended to monitor progress.

```python
import logging

project_number = PROJECT_NUMBER # replace with your project number
service_account = f"{project_number}-compute@developer.gserviceaccount.com"

vapo_config = types.PromptOptimizerVAPOConfig(
    config_path="gs://your-bucket/config.json",
    service_account_project_number=project_number,
    wait_for_completion=False
)

# Set up logging to see the progress of the optimization job
logging.basicConfig(encoding='utf-8', level=logging.INFO, force=True)

result = client.prompt_optimizer.optimize(method="vapo", config=vapo_config)
```

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

### Get Predictions from Endpoint

Source: https://github.com/googleapis/python-aiplatform/blob/main/docs/README.rst

Sends prediction requests to a deployed model on a Vertex AI Endpoint. Accepts a list of instances for prediction.

```Python
endpoint.predict(instances=[[6.7, 3.1, 4.7, 1.5], [4.6, 3.1, 1.5, 0.2]])
```

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

### Select a Verified Deployment: By Container Image

Source: https://github.com/googleapis/python-aiplatform/blob/main/agentplatform/model_garden/README.md

Deploys a model using a specified serving container image URI from the verified deployment configurations.

```APIDOC
## Select a Verified Deployment: By Container Image

Specify a container image from the list of verified deployment configurations.

### Code Example
```python
endpoint = model.deploy(
    serving_container_image_uri="us-docker.pkg.dev/vertex-ai/vertex-vision-model-garden-dockers/pytorch-vllm-serve:20250430_0916_RC00_maas",
)
```
```

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

### Initialize Vertex AI SDK

Source: https://github.com/googleapis/python-aiplatform/blob/main/gemini_docs/README.md

Initializes the Vertex AI SDK with your project ID and location. This must be done before using any other SDK functionalities.

```python
vertexai.init(project='my-project', location='us-central1')
```

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

### Get TensorFlow 2 Explainable AI Metadata

Source: https://github.com/googleapis/python-aiplatform/blob/main/README.rst

Retrieve metadata for TensorFlow 2 models in dictionary format. Uses the `google.cloud.aiplatform.explain.metadata.tf.v2` module.

```Python
from google.cloud.aiplatform.explain.metadata.tf.v2 import saved_model_metadata_builder

builder = saved_model_metadata_builder.SavedModelMetadataBuilder('gs://python/to/my/model/dir')
generated_md = builder.get_metadata()
```

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

### List Model Deployment Configurations

Source: https://github.com/googleapis/python-aiplatform/blob/main/vertexai/model_garden/README.md

Inspect the available deployment configurations, such as compatible machine types and container images, for a specific model.

```python
model = model_garden.OpenModel("google/paligemma@paligemma-224-float32")
deploy_options = model.list_deploy_options()
```

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

### Get TensorFlow 1 Explainable AI Metadata

Source: https://github.com/googleapis/python-aiplatform/blob/main/README.rst

Retrieve metadata for TensorFlow 1 models in dictionary format. Requires the `google.cloud.aiplatform.explain.metadata.tf.v1` module.

```Python
from google.cloud.aiplatform.explain.metadata.tf.v1 import saved_model_metadata_builder

builder = saved_model_metadata_builder.SavedModelMetadataBuilder(
          'gs://python/to/my/model/dir', tags=[tf.saved_model.tag_constants.SERVING]
      )
generated_md = builder.get_metadata()
```

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

### Deploy Agent to Agent Engine

Source: https://github.com/googleapis/python-aiplatform/blob/main/gemini_docs/README.md

Initialize Vertex AI and create a remote agent instance. This prepares the agent for deployment to Agent Engine.

```python
vertexai.init(
    project='my-project',
    location='us-central1',
    staging_bucket="gs://my-staging-bucket",
)

remote_app = vertexai.agent_engines.create(
    app,
    requirements=["google-cloud-aiplatform[agent_engines,adk]"],
)
```

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

### Instantiate Vertex AI Client

Source: https://github.com/googleapis/python-aiplatform/blob/main/README.rst

Instantiate a Vertex AI client for generative AI features. Replace 'my-project' and 'us-central1' with your project ID and location.

```Python
import vertexai
from vertexai import types

# Instantiate GenAI client from Vertex SDK
# Replace with your project ID and location
client = vertexai.Client(project='my-project', location='us-central1')
```

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

### Update Skill Metadata

Source: https://github.com/googleapis/python-aiplatform/blob/main/README.rst

Updates an existing skill's metadata or underlying implementation. This example shows updating the display name and description.

```Python
# Update skill metadata
updated_skill = client.skills.update(
    name=skill.name,
    config={
        "display_name": "Updated Weather Skill",
        "description": "Provides localized current weather conditions and multi-day forecasts.",
    },
)
```

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

### Test Agent Locally

Source: https://github.com/googleapis/python-aiplatform/blob/main/gemini_docs/README.md

Test the locally defined ADK Agent by streaming queries. This example queries the exchange rate from USD to SEK.

```python
async for event in app.async_stream_query(
    user_id="user-id",
    message="What is the exchange rate from US dollars to SEK today?",
):
    print(event)
```

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

### Run All System Tests

Source: https://github.com/googleapis/python-aiplatform/blob/main/CONTRIBUTING.rst

Execute all system tests for the project using the nox 'system' session.

```bash
nox -s system
```

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

### Create Gen AI Client

Source: https://github.com/googleapis/python-aiplatform/blob/main/README.rst

Initializes a client for the Gen AI SDK, specifying Vertex AI integration, project, and location.

```Python
# Create a Client in the Gen AI SDK
genai_client = genai.Client(vertexai=True, project="your-project", location="your-location")

# Call generate_content() with the prompt
response = genai_client.models.generate_content(
    model=retrieved_prompt.prompt_data.model,
    contents=retrieved_prompt.assemble_contents(),
)
```

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

### Read Training Data URIs

Source: https://github.com/googleapis/python-aiplatform/blob/main/README.rst

Access environment variables to get the format and URIs for training, validation, and test data splits during custom training.

```Python
os.environ['AIP_DATA_FORMAT']  # provides format of data
os.environ['AIP_TRAINING_DATA_URI']  # uri to training split
os.environ['AIP_VALIDATION_DATA_URI']  # uri to validation split
os.environ['AIP_TEST_DATA_URI']  # uri to test split
```

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

### Deploy Model with Explanation Metadata

Source: https://github.com/googleapis/python-aiplatform/blob/main/docs/README.rst

Use explanation metadata when deploying a model to an endpoint or uploading a model to Vertex AI. This enables explainability features for predictions.

```Python
explanation_metadata = builder.get_metadata_protobuf()

# To deploy a model to an endpoint with explanation
model.deploy(..., explanation_metadata=explanation_metadata)

# To deploy a model to a created endpoint with explanation
endpoint.deploy(..., explanation_metadata=explanation_metadata)

# To upload a model with explanation
aiplatform.Model.upload(..., explanation_metadata=explanation_metadata)
```

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

### Get a Model Evaluation

Source: https://github.com/googleapis/python-aiplatform/blob/main/docs/README.rst

Retrieves a specific model evaluation resource for a given Vertex AI Model. Can fetch the first evaluation or one by its ID.

```Python
model = aiplatform.Model('projects/my-project/locations/us-central1/models/{MODEL_ID}')

# returns the first evaluation with no arguments, you can also pass the evaluation ID
evaluation = model.get_model_evaluation()

eval_metrics = evaluation.metrics
```

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

### Clone Repository

Source: https://github.com/googleapis/python-aiplatform/blob/main/scripts/readme-gen/templates/install_deps.tmpl.rst

Clone the python-docs-samples repository from GitHub to your local machine.

```bash
$ git clone https://github.com/GoogleCloudPlatform/python-docs-samples.git
```

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

### Generate Content with Prompt (Gen AI SDK)

Source: https://github.com/googleapis/python-aiplatform/blob/main/gemini_docs/README.md

After creating or retrieving a prompt, use the Gen AI SDK's `generate_content` method. The `assemble_contents()` utility function on Prompt objects converts the prompt into a format suitable for `generate_content`.

```python
from google import genai
from google.genai import types as genai_types

# Create a Client in the Gen AI SDK
genai_client = genai.Client(vertexai=True, project="your-project", location="your-location")

# Call generate_content() with the prompt
response = genai_client.models.generate_content(
    model=retrieved_prompt.prompt_data.model,
    contents=retrieved_prompt.assemble_contents(),
)
```

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

### Get Serving Container Specification

Source: https://github.com/googleapis/python-aiplatform/blob/main/google/cloud/aiplatform/prediction/README.md

Retrieve the serving container specification for the built image using `local_model.get_serving_container_spec()`. This is useful for understanding the container's configuration.

```python
local_model.get_serving_container_spec()
```

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

### Spot VM Deployment

Source: https://github.com/googleapis/python-aiplatform/blob/main/agentplatform/model_garden/README.md

Schedules workloads on Spot VMs for lower cost by setting `spot=True` during deployment.

```APIDOC
## Spot VM Deployment

Schedule workloads on Spot VMs for lower cost.

### Code Example
```python
endpoint = model.deploy(spot=True)
```

**Use case:** Cost-sensitive development and batch workloads.
```

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

### Get Existing Image Dataset

Source: https://github.com/googleapis/python-aiplatform/blob/main/docs/README.rst

Retrieve a previously created image dataset using its resource name. Replace '{DATASET_ID}' with the actual dataset ID.

```Python
dataset = aiplatform.ImageDataset('projects/my-project/location/us-central1/datasets/{DATASET_ID}')
```

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

### Deploy Model with EULA Acceptance

Source: https://github.com/googleapis/python-aiplatform/blob/main/agentplatform/model_garden/README.md

Deploy models that require license agreement acceptance by passing `eula=True`. Use this for the first-time deployment of EULA-protected models.

```python
model = model_garden.OpenModel("google/gemma2@gemma-2-27b-it")
endpoint = model.deploy(eula=True)
```

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

### Get Existing Image Dataset

Source: https://github.com/googleapis/python-aiplatform/blob/main/README.rst

Retrieve a previously created image dataset using its full resource name, including project, location, and dataset ID.

```Python
dataset = aiplatform.ImageDataset('projects/my-project/location/us-central1/datasets/{DATASET_ID}')
```

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

### Get a Vertex AI Model

Source: https://github.com/googleapis/python-aiplatform/blob/main/docs/README.rst

Retrieves a Vertex AI Model resource using its resource name. This is useful for subsequent operations like deployment or evaluation.

```Python
model = aiplatform.Model('/projects/my-project/locations/us-central1/models/{MODEL_ID}')
```

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

### Data-Driven Prompt Optimization with VAPOConfig

Source: https://github.com/googleapis/python-aiplatform/blob/main/README.rst

Launch a data-driven prompt optimization job using VAPOConfig. This requires a project number and a GCS path to the configuration file. Ensure logging is set up for progress monitoring.

```Python
from vertexai import types
import logging

project_number = PROJECT_NUMBER # replace with your project number
service_account = f"{project_number}-compute@developer.gserviceaccount.com"

vapo_config = vertexai.types.PromptOptimizerVAPOConfig(
    config_path="gs://your-bucket/config.json",
    service_account_project_number=project_number,
    wait_for_completion=False
)

# Set up logging to see the progress of the optimization job
logging.basicConfig(encoding='utf-8', level=logging.INFO, force=True)

result = client.prompts.launch_optimization_job(method=types.PromptOptimizerMethod.VAPO, config=vapo_config)
```

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

### Create Pipeline Run

Source: https://github.com/googleapis/python-aiplatform/blob/main/README.rst

Instantiate and run a Vertex AI PipelineJob. Configure display name and caching behavior.

```Python
# Instantiate PipelineJob object
pl = PipelineJob(
  display_name="My first pipeline",

  # Whether or not to enable caching
  # True = always cache pipeline step result
)
```

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

### Get TensorFlow 2 Explainable AI Metadata

Source: https://github.com/googleapis/python-aiplatform/blob/main/docs/README.rst

Retrieve metadata in dictionary format from TensorFlow 2 models using SavedModelMetadataBuilder. This version does not require specific tags.

```Python
from google.cloud.aiplatform.explain.metadata.tf.v2 import saved_model_metadata_builder

builder = saved_model_metadata_builder.SavedModelMetadataBuilder('gs://python/to/my/model/dir')
generated_md = builder.get_metadata()
```

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

### Get TensorFlow 1 Explainable AI Metadata

Source: https://github.com/googleapis/python-aiplatform/blob/main/docs/README.rst

Retrieve metadata in dictionary format from TensorFlow 1 models using SavedModelMetadataBuilder. Ensure the correct tags are provided.

```Python
from google.cloud.aiplatform.explain.metadata.tf.v1 import saved_model_metadata_builder

builder = saved_model_metadata_builder.SavedModelMetadataBuilder(
          'gs://python/to/my/model/dir', tags=[tf.saved_model.tag_constants.SERVING]
      )
generated_md = builder.get_metadata()
```

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

### Clone and Set Up Local Repository

Source: https://github.com/googleapis/python-aiplatform/blob/main/CONTRIBUTING.rst

Clone the python-aiplatform repository from your GitHub fork and configure remotes to pull changes from the upstream repository.

```bash
cd ${HOME}
git clone git@github.com:USERNAME/python-aiplatform.git hack-on-python-aiplatform
cd hack-on-python-aiplatform
# Configure remotes such that you can pull changes from the googleapis/python-aiplatform
# repository into your local repository.
git remote add upstream git@github.com:googleapis/python-aiplatform.git
# fetch and merge changes from upstream into main
git fetch upstream
git merge upstream/main
```

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

### Initialize Vertex AI Client

Source: https://github.com/googleapis/python-aiplatform/blob/main/gemini_docs/README.md

Initialize the Vertex AI client with your project ID and desired location. This client is used to interact with Vertex AI services.

```python
client = vertexai.Client(project='my-project', location='us-central1')
```

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

### Deploy Hugging Face Model

Source: https://github.com/googleapis/python-aiplatform/blob/main/vertexai/model_garden/README.md

Deploy a model directly from Hugging Face using its model ID. This is useful for leveraging community models without custom container setup.

```python
model = model_garden.OpenModel("Qwen/Qwen2-1.5B-Instruct")
endpoint = model.deploy()
```

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

### Create and Import Text Dataset

Source: https://github.com/googleapis/python-aiplatform/blob/main/docs/README.rst

Create a text dataset and then import data from a GCS source. Ensure the import schema URI is correctly specified for the dataset type.

```Python
from google.cloud import aiplatform

my_dataset = aiplatform.TextDataset.create(
    display_name="my-dataset")

my_dataset.import_data(
    gcs_source=['gs://path/to/my/dataset.csv'],
    import_schema_uri=aiplatform.schema.dataset.ioformat.text.multi_label_classification
)
```

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

### Define Function for Model Use

Source: https://github.com/googleapis/python-aiplatform/blob/main/gemini_docs/README.md

Defines a function declaration, including its schema, name, description, and parameters, that a generative model can use to answer questions. This example defines a 'get_current_weather' function.

```python
# First, create tools that the model is can use to answer your questions.
# Describe a function by specifying it's schema (JsonSchema format)
get_current_weather_func = generative_models.FunctionDeclaration(
    name="get_current_weather",
    description="Get the current weather in a given location",
    parameters={
        "type": "object",
        "properties": {
            "location": {
                "type": "string",
                "description": "The city and state, e.g. San Francisco, CA"
            },
            "unit": {
                "type": "string",
                "enum": [
                    "celsius",
                    "fahrenheit",
                ]
            }
        },
        "required": [
            "location"
        ]
    },
)
# Tool is a collection of related functions
weather_tool = generative_models.Tool(
    function_declarations=[get_current_weather_func],
)
```

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

### Deploy Model with Specific Hardware

Source: https://github.com/googleapis/python-aiplatform/blob/main/agentplatform/model_garden/README.md

Deploy a model using a specified machine type, accelerator type, and count from the verified deployment configurations.

```python
endpoints = model.deploy(
    machine_type="a3-highgpu-1g",
    accelerator_type="NVIDIA_H100_80GB",
    accelerator_count=1,
)
```

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

### Create Skill

Source: https://github.com/googleapis/python-aiplatform/blob/main/README.rst

Demonstrates how to create a new skill in the Skill Registry with optional custom ID.

```APIDOC
## Create Skill

### Description
Creates a new skill in the Skill Registry. You can optionally specify a custom string identifier using the `skill_id` configuration parameter.

### Method
`client.skills.create()`

### Parameters
- **display_name** (str) - Required - The display name for the skill.
- **description** (str) - Required - A description of the skill's functionality.
- **config** (dict) - Optional - Configuration for the skill, including `local_path` and `skill_id`.
  - **local_path** (str) - The local path to the skill's implementation.
  - **skill_id** (str) - A custom string identifier for the skill.

### Request Example
```python
skill = client.skills.create(
    display_name="weather_skill",
    description="Retrieves the weather for a given location",
    config={
        "local_path": "./weather_skill_dir",
        "skill_id": "my-custom-weather-skill",
    },
)
```
```

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

### Get Existing Skill by Resource Name

Source: https://github.com/googleapis/python-aiplatform/blob/main/gemini_docs/README.md

Retrieve an existing skill using its full resource name, which follows the format 'projects/{project_id_or_number}/locations/{location}/skills/{skill_id}'. The `skill.name` attribute from a `create()` call provides this format directly.

```python
# The `name` parameter must be a string in the following format:
#   'projects/{project_id_or_number}/locations/{location}/skills/{skill_id}'
#
# Examples:
#   'projects/my-project/locations/us-central1/skills/1234567890'
#   'projects/my-project/locations/us-central1/skills/my-custom-weather-skill'
#
# Note: If you have a `Skill` object (e.g., from a `create()` call),
# `skill.name` already contains the resource name in the correct format.
fetched_skill = client.skills.get(name=skill.name)
```