anthropic-vertex-ai-provider v1.0.2

Vercel AI SDK - Anthropic Vertex Provider

fry69/anthropic-vertex-ai-provider is a community provider that uses Anthropic models through Vertex AI to provide language model support for the Vercel AI SDK.

This provider is a community-maintained module and is not officially supported by Vercel. It was forked from the Anthropic Vertex Provider pull request and updated.

Compatibility

The AnthropicVertex provider was tested with Vercel AI SDK version 3.1.0. To avoid TypeScript errors it is recommended to use version 3.3.0 or higher.

Example scripts

Example scripts how to use this provider can be found in the examples folder.

Setup

The AnthropicVertex provider is available in the anthropic-vertex-ai-provider module. You can install it with:

npm install anthropic-vertex-ai-provider
# or
pnpm add anthropic-vertex-ai-provider
# or
yarn add anthropic-vertex-ai-provider
# or
bun add anthropic-vertex-ai-provider

Provider Instance

You can import the default provider instance anthropicVertex from anthropic-vertex-ai-provider:

import { anthropicVertex } from 'anthropic-vertex-ai-provider';

If you need a customized setup, you can import createAnthropicVertex from anthropic-vertex-ai-provider and create a provider instance with your settings:

import { createAnthropicVertex } from 'anthropic-vertex-ai-provider';

const anthropicVertex = createAnthropicVertex({
  region: 'us-central1',
  projectId: 'your-project-id',
  // other options
});

You can use the following optional settings to customize the AnthropicVertex provider instance:

• region string

  Your Google Vertex region. Defaults to the GOOGLE_VERTEX_REGION environment variable.

• projectId string

  Your Google Vertex project ID. Defaults to the GOOGLE_VERTEX_PROJECT_ID environment variable.

• googleAuth GoogleAuth

  Optional. The Authentication options provided by google-auth-library.

• baseURL string

  Use a different URL prefix for API calls, e.g., to use proxy servers. The default prefix is https://{region}-aiplatform.googleapis.com/v1.

• headers Record<string,string>

  Custom headers to include in the requests.

• fetch (input: RequestInfo, init?: RequestInit) => Promise<Response>

  Custom fetch implementation. You can use it as a middleware to intercept requests, or to provide a custom fetch implementation for e.g., testing.

Language Models

You can create models that call the Anthropic API through Vertex AI using the provider instance. The first argument is the model ID, e.g., claude-3-sonnet@20240229:

const model = anthropicVertex('claude-3-sonnet@20240229');

Example: Generate Text

You can use AnthropicVertex language models to generate text with the generateText function:

import { anthropicVertex } from 'anthropic-vertex-ai-provider';
import { generateText } from 'ai';

const { text } = await generateText({
  model: anthropicVertex('claude-3-sonnet@20240229'),
  prompt: 'Write a vegetarian lasagna recipe for 4 people.',
});

AnthropicVertex language models can also be used in the streamText, generateObject, and streamObject functions (see AI SDK Core for more information).

Model Capabilities

Environment Variables

To use the AnthropicVertex provider, you need to set up the following environment variables:

• GOOGLE_VERTEX_REGION: Your Google Vertex region (e.g., 'us-central1')
• GOOGLE_VERTEX_PROJECT_ID: Your Google Cloud project ID

Make sure to set these variables in your environment or in a .env file in your project root.

Authentication

The AnthropicVertex provider uses Google Cloud authentication. Make sure you have set up your Google Cloud credentials properly. You can either use a service account key file or default application credentials.

For more information on setting up authentication, refer to the Google Cloud Authentication guide.

Tests

To run the tests, make sure you have installed the dependencies with:

pnpm install

Then, run the tests with:

pnpm test

All tests will get run automatically on every push to the repository to ensure regressions will get caught as early as possible.