Skip to main content

Module llm

Module llm

Expand description

LLM abstraction layer for multiple providers

Structs§

AssistantMessage: Assistant message
CacheControl: Cache control for prompt caching
ChatAnthropic: Anthropic Chat Model
ChatCompletion: Chat completion response
ChatDeepSeek: DeepSeek Chat Model
ChatGoogle: Google Gemini Chat Model
ChatGroq: Groq Chat Model
ChatMistral: Mistral Chat Model
ChatOllama: Ollama Chat Model
ChatOpenAI: OpenAI Chat Model
ChatOpenAICompatible: OpenAI-compatible Chat Model base implementation
ChatOpenRouter: OpenRouter Chat Model
ContentPartDocument: Document content part (for PDFs etc.)
ContentPartImage: Image content part
ContentPartRedactedThinking: Redacted thinking content
ContentPartRefusal: Refusal content part
ContentPartText: Text content part
ContentPartThinking: Thinking content part
DeveloperMessage: Developer message (for o1+ models)
DocumentSource
Function: Function call from the LLM
ImageUrl: Image URL structure
ModelBuilder: Builder pattern helpers for common model configurations
SchemaOptimizer: Schema optimizer for creating LLM-compatible JSON schemas
SystemMessage: System message
ToolCall: Tool call from the LLM
ToolDefinition: Definition of a tool that can be called by the LLM
ToolMessage: Tool result message
Usage: Token usage information
UserMessage: User message

Enums§

CacheControlType
ContentPart: Union type for all content parts
LlmError: Error types for LLM operations
Message: Union type for all messages
ReasoningEffort: Reasoning effort levels for o1+ models
StopReason: Stop reason for completion
ToolChoice: Tool choice strategy

Traits§

BaseChatModel: Base trait for chat model implementations

Type Aliases§

ChatStream: Type alias for boxed stream
JsonSchema: JSON Schema for tool parameters