Vercel AI SDK

Vercel AI Provider

Use ArtemisKit with the Vercel AI SDK, which provides a unified interface to multiple LLM providers.

Supported Underlying Providers

The Vercel AI adapter currently supports:

Provider	Status
OpenAI	✅ Supported
Azure OpenAI	✅ Supported
Anthropic	🔜 Coming soon
Google	🔜 Coming soon
Mistral	🔜 Coming soon

Setup

1. Install Dependencies

The Vercel AI SDK is included with ArtemisKit. No additional installation required.

2. Set Environment Variables

For OpenAI:

export OPENAI_API_KEY="sk-..."

For Azure OpenAI:

export AZURE_OPENAI_API_KEY="your-key"
export AZURE_OPENAI_RESOURCE_NAME="your-resource"

3. Configure Scenario

name: my-test
provider: vercel-ai
model: gpt-5

providerConfig:
  underlyingProvider: openai  # or 'azure'

cases:
  - id: example
    prompt: "Hello"
    expected:
      type: contains
      values: ["hello"]
      mode: any

Config File

Configure Vercel AI in your artemis.config.yaml:

provider: vercel-ai
model: gpt-5

providers:
  vercel-ai:
    apiKey: ${OPENAI_API_KEY}
    underlyingProvider: openai
    timeout: 60000

Azure Configuration

provider: vercel-ai
model: gpt-5

providers:
  vercel-ai:
    apiKey: ${AZURE_OPENAI_API_KEY}
    underlyingProvider: azure
    providerConfig:
      resourceName: ${AZURE_OPENAI_RESOURCE_NAME}

Why Use Vercel AI?

Unified API — Same interface across different providers
Easy switching — Change providers with minimal code changes
Future-proof — Access to new providers as they’re added

Available Models

Models depend on the underlying provider:

OpenAI (underlyingProvider: openai)

Model	Description
`gpt-5`	Latest flagship model
`gpt-4o`	Multimodal, versatile
`gpt-4o-mini`	Fast and cost-effective

Azure (underlyingProvider: azure)

Use your Azure deployment name as the model. The actual model depends on your Azure deployment configuration.