Skip to main content

Providers and Credentials

This page covers how to think about provider configuration in Spring AI: selecting providers, managing credentials, and handling provider-specific capabilities without leaking them into your core business logic.

What You Configure​

  • Provider selection (OpenAI, Anthropic, Azure OpenAI, Ollama, etc.).
  • Model selection and per-model constraints (context length, tool calling support, pricing).
  • Runtime options (temperature, top-p, max tokens) as product defaults, not ad-hoc constants.
  • Credential management (environment variables, secret managers, rotation, least privilege).

Trade-offs​

  • Single provider is simpler, but increases vendor risk.
  • Multi-provider improves resilience and cost control, but increases testing and compatibility work.

Coming Next​

  • A provider capability matrix and a “portable baseline” configuration strategy.
  • Patterns for provider failover and graceful degradation.