Skip to main content
You can generate text completions using any supported provider through the Prism Vertex package. The provider uses the Prism text() method with the Vertex provider constants.

​
Basic usage

Generate a text response using the Prism::text() method with a Vertex provider constant: 
use Prism\Prism\Prism;
use Prism\Vertex\Enums\Vertex;

$response = Prism::text()
    ->using(Vertex::Gemini, 'gemini-2.5-flash')
    ->withPrompt('Explain quantum computing in simple terms')
    ->asText();

echo $response->text;
The using() method accepts two parameters:
  • Provider constant - One of the Vertex::* constants (e.g., Vertex::Gemini, Vertex::Anthropic)
  • Model string - The specific model identifier

​
Supported providers

Prism Vertex supports multiple AI providers through a single configuration: 
$response = Prism::text()
    ->using(Vertex::Gemini, 'gemini-2.5-flash')
    ->withPrompt('Explain quantum computing in simple terms')
    ->asText();

​
Available provider constants

All provider constants are available in the Prism\Vertex\Enums\Vertex class:
ConstantPublisherExample Models
Vertex::Geminigooglegemini-2.5-flash, gemini-pro
Vertex::Anthropicanthropicclaude-3-5-sonnet@20241022, claude-3-5-haiku@20241022
Vertex::Mistralmistralaimistral-small-2503, codestral-2501
Vertex::Metametallama-4-scout-17b-16e-instruct-maas
Vertex::DeepSeekdeepseekdeepseek-v3-0324-maas
Vertex::AI21ai21jamba-1.5-mini@001, jamba-1.5-large@001
Vertex::Kimikimikimi-k2-0711-maas
Vertex::MiniMaxminimaxminimax-m1-40k-0709-maas
Vertex::OpenAIopenaigpt-oss-4o-mini-maas
Vertex::Qwenqwenqwen2.5-72b-instruct-maas
Vertex::ZAIzaiorgglm-4-plus-maas

​
Response object

The asText() method returns a Prism\Prism\Text\Response object with the following properties: 
$response = Prism::text()
    ->using(Vertex::Gemini, 'gemini-2.5-flash')
    ->withPrompt('Hello, world!')
    ->asText();

// Access the generated text
echo $response->text;

// Access usage statistics (when available)
echo $response->usage->inputTokens;
echo $response->usage->outputTokens;

// Access the raw response
var_dump($response->response);

​
API schema selection

Prism Vertex automatically selects the correct API schema based on the provider constant and model string. Three schemas are supported:
The schema is selected automatically - you don’t need to specify it unless you want to override the default behavior.

​
Gemini schema

Used for Google Gemini models. Uses the native generateContent endpoint. 
$response = Prism::text()
    ->using(Vertex::Gemini, 'gemini-2.5-flash')
    ->withPrompt('Hello')
    ->asText();

​
Anthropic schema

Used for Anthropic Claude models. Uses the :rawPredict endpoint with Anthropic Messages API format. 
$response = Prism::text()
    ->using(Vertex::Anthropic, 'claude-3-5-sonnet@20241022')
    ->withPrompt('Hello')
    ->asText();

​
OpenAI schema

Used for partner models (Mistral, Meta, DeepSeek, etc.). Uses :rawPredict or :chatCompletions with OpenAI-compatible format. 
$response = Prism::text()
    ->using(Vertex::Mistral, 'mistral-small-2503')
    ->withPrompt('Hello')
    ->asText();

​
Overriding the API schema

You can override the automatically selected schema using withProviderOptions(): 
use Prism\Vertex\Enums\VertexSchema;

$response = Prism::text()
    ->using(Vertex::Gemini, 'some-model')
    ->withProviderOptions(['apiSchema' => VertexSchema::Anthropic])
    ->withPrompt('Hello')
    ->asText();
Overriding the schema is an advanced feature. The automatic selection works for all standard use cases.

​
Express mode vs Standard mode

Prism Vertex supports two authentication modes:

​
Express mode (API key only)

When project_id and location are omitted, the package automatically uses Vertex AI Express Mode endpoints:
config/prism.php
'vertex' => [
    'api_key' => env('VERTEX_API_KEY'),
    // project_id and location omitted — triggers Express mode
],
Express mode only supports Google Gemini models. Partner models require Standard mode.

​
Standard mode (project + location)

Standard mode supports all providers and authentication methods:
config/prism.php
'vertex' => [
    'project_id'  => env('VERTEX_PROJECT_ID'),
    'location'    => env('VERTEX_LOCATION', 'us-central1'),
    'credentials' => env('VERTEX_CREDENTIALS'), // path to service-account.json
    // OR use api_key instead of credentials:
    // 'api_key' => env('VERTEX_API_KEY'),
],

​
Error handling

The package handles Vertex AI errors and converts them to Prism exceptions: 
use Prism\Prism\Exceptions\PrismException;
use Prism\Prism\Exceptions\PrismRateLimitedException;
use Prism\Prism\Exceptions\PrismProviderOverloadedException;

try {
    $response = Prism::text()
        ->using(Vertex::Gemini, 'gemini-2.5-flash')
        ->withPrompt('Hello')
        ->asText();
} catch (PrismRateLimitedException $e) {
    // Handle rate limiting (HTTP 429)
    echo 'Rate limited: ' . $e->getMessage();
} catch (PrismProviderOverloadedException $e) {
    // Handle provider overload (HTTP 503, 529)
    echo 'Provider overloaded: ' . $e->getMessage();
} catch (PrismException $e) {
    // Handle other errors
    echo 'Error: ' . $e->getMessage();
}

​
Next steps

Structured output

Generate JSON responses with schema validation

Embeddings

Create text embeddings for semantic search

Multi-provider

Use multiple providers with different configurations

Build docs developers (and LLMs) love

Get started for freeTalk to us