Model Configuration

By default, Puck AI uses a managed model charged on a pay-as-you-go basis.

In our Launch plan and above, you can control which model is used, bring your own provider key, and fine-tune the settings.

See model availability for supported models.

The managed model

Puck’s managed model provides great out-of-the-box behavior for Puck AI, charged on a pay-as-you-go, token consumption basis:

You can top-up your Puck Cloud account any time, or specify auto-reload to ensure you never run out of credit.

Specify a model

To specify a different model, set the model and providerApiKey parameters in the puckHandler:

const handler = puckHandler(request, {
  ai: {
    model: "openai/gpt-5.5",
    providerApiKey: process.env.MY_OPENAI_KEY, // Must be specified. Will default to OPENAI_API_KEY
  },
});

You can also bring your own key in the generate and chat APIs.

Different providers impose rate limits on new accounts. If you encounter these, try to increase your limit with the provider or use our managed model.

Controlling model parameters

Use the providerOptions parameter to fine-tune model behavior, such as to control reasoning effort:

const handler = puckHandler(request, {
  ai: {
    model: "openai/gpt-5.5",
    providerApiKey: process.env.MY_OPENAI_KEY,
    providerOptions: {
      openai: {
        reasoningEffort: "low",
      },
    },
  },
});

Model availability

We currently support Open AI models, and are working to expand this.

Further reading

• puckHandler API reference
• generate API reference
• chat API reference