Skip to content

Configuration

All settings live at /admin/config/ai/providers/dxpr. Permission required: Administer AI providers.

API key

The API key is managed through the Drupal Key module. The settings form provides a key selector that lists all Key entities on your site. On initial install, the module auto-detects any Key entity with "DXPR" in the label and pre-selects it.

DXPR AI Provider models list with failover configuration

Model selection

Three model variants are available: kavya-m1 (general use), kavya-m1-eu (EU data residency), and kavya-m1-fast (low latency). All three support chat (including vision, tool calls, and structured JSON output), streaming, and translation. See the model table for default providers and fallback chains.

Provider selection mode

Automatic (default) uses Kavya's built-in provider priority. Manual Override shows a drag-and-drop table where you control the failover order. See the failover guide for details.

Provider priority table

Visible when Manual Override is selected. The drag-and-drop table shows five upstream providers: OpenAI, Anthropic, Google Gemini, MistralAI, and XAI. Each can be:

  • Enabled: included in the failover chain at its configured priority
  • Disabled: excluded from routing entirely

The top-most enabled provider handles requests first. If it fails, the platform tries the next enabled provider in order. Each row shows the provider's logo and country of origin for data residency awareness.

See the multi-provider failover guide for details on how failover works and a comparison with other Drupal AI providers.

AI usage tracking

The settings page displays your current credit balance and monthly usage, pulled from the DXPR licence service. Credits are shared across all AI operations:

  • 1 credit = 1 generated word
  • 1 image generation = 1,000 credits
  • HTML markup does not count toward usage

Em-dash processor

The em-dash post-processing engine removes AI-generated em-dashes from output text. It is configured with a slider on the settings form. See the em-dash post-processing guide for full details on the four intensity levels and per-language overrides.

Next steps