Vercel AI Provider

Vercel AI is a gateway that provides unified access to multiple AI providers. Use it when you have an existing Vercel AI setup or want to route through Vercel’s infrastructure.

Prerequisites

1. A Vercel account with AI access
2. Your Vercel AI gateway URL
3. An API key for the underlying provider

Configure VibeCody

Config file (~/.vibecli/config.toml)

[vercel_ai]
enabled = true
api_key = "your-provider-api-key"
api_url = "https://your-vercel-ai-endpoint.vercel.app/api"
model = "gpt-4o"

Both api_key and api_url are required. The API URL points to your Vercel AI gateway endpoint.

Model Selection

Available models depend on what your Vercel AI gateway is configured to serve.

Default: gpt-4o

Override from the CLI:

vibecli --provider vercel_ai --model gpt-4o-mini

Best For

• Existing Vercel infrastructure – route AI through your Vercel deployment
• Custom middleware – add logging, caching, or rate limiting via Vercel functions
• Multi-provider gateway – use Vercel AI as a proxy to multiple backends

Verify Connection

vibecli --provider vercel_ai -c "Hello, which model are you?"

Troubleshooting

Connection refused

• Verify api_url points to a running Vercel AI endpoint
• Check that the endpoint supports the OpenAI-compatible chat completions format

Authentication error

• The api_key is passed to the underlying provider through Vercel – ensure it is valid for that provider