LLM Scope Levels
- System LLM: Default global model used when no workspace or agent-specific model is defined. Supports multi-modal input.
- Workspace LLM: Workspace-specific override. Applies only when chatting within that specific workspace. Supports multi-modal input.
- Agent LLM: Dedicated model configuration for AI agents. Recommended to use models with native tool-calling support.
Provider Types
- Local Providers: Host and run LLMs locally.
- Cloud Providers: Interface with external cloud-based LLM APIs.