Model Provider Configurationfastbee Less than 1 minute Model providers store the connection information for external or local large-model services. A provider is usually configured once, while one or more models can be attached to it.
Field Description Recommendation Provider name Business-readable name. Include vendor and environment. Endpoint Model API endpoint. Use intranet endpoint for private deployment. API key Authentication credential. Store securely and rotate regularly. Status Whether the provider is available. Disable unused providers. Timeout Request timeout. Set according to model latency.
Field Description Model name Name used by the API or local model service. Model type Chat, embedding, rerank, or other capability. Context length Maximum input context supported by the model. Temperature Creativity and stability parameter. Enabled status Whether the model can be selected by business features.
Item Expected Result Provider connectivity The model endpoint can be reached from the backend server. Key validation Invalid keys return a clear error and valid keys pass. Chat test A simple question returns a normal answer. Knowledge retrieval Embedding or retrieval model works when RAG is enabled. Error handling Timeout and quota errors are logged clearly.
Prev
AI Overview
Next
Knowledge Base And RAG