dartantic_ai library

Classes

Agent: An agent that manages chat models and provides tool execution and message collection capabilities.
AnthropicChatModel: Wrapper around Anthropic Messages API (aka Claude API).
AnthropicChatOptions: Options to pass into the Anthropic Chat Model.
AnthropicMediaGenerationModel: Media generation model backed by the Anthropic code execution tool.
AnthropicMediaGenerationModelOptions: Options for configuring Anthropic media generation runs.
AnthropicProvider: Provider for Anthropic Claude native API.
AnthropicServerToolConfig: Configuration for enabling Anthropic server-side tools.
AnthropicToolChoice: Determines how Claude should select server-side tools.
BatchEmbeddingsResult: Result for batch embeddings operations.
Chat: A chat session with an agent.
ChatGoogleGenerativeAISafetySetting: Safety setting, affecting the safety-blocking behavior. Passing a safety setting for a category changes the allowed probability that content is blocked.
ChatMessage: A message in a conversation between a user and a model.
ChatModel<TOptions extends ChatModelOptions>: Chat model base class.
ChatModelOptions: Generation options to pass into the Chat Model.
ChatResult<T extends Object>: Result returned by the Chat Model.
CodeInterpreterConfig: Configuration for the OpenAI Responses code_interpreter tool.
CohereEmbeddingsModel: Cohere embeddings model implementation.
CohereEmbeddingsModelOptions: Options for Cohere embeddings models.
CohereProvider: Provider for Cohere OpenAI-compatible API.
ContainerFileData: Resolved data for a downloaded container file, including metadata hints.
DataPart: A data part containing binary data (e.g., images).
DefaultStreamingOrchestrator: Default implementation of the streaming orchestrator.
EmbeddingsModel<TOptions extends EmbeddingsModelOptions>: Embeddings model base class.
EmbeddingsModelOptions: Base class for embeddings model options.
EmbeddingsResult: Result returned by embeddings providers.
FileSearchConfig: Configuration for the OpenAI Responses file_search tool.
GoogleChatModel: Wrapper around Google AI for Developers API (aka Gemini API).
GoogleChatModelOptions: Options to pass into the Google Generative AI Chat Model.
GoogleEmbeddingsModel: Google AI embeddings model implementation.
GoogleEmbeddingsModelOptions: Google AI-specific embeddings model options.
GoogleMediaGenerationModel: Media generation model for Google Gemini.
GoogleMediaGenerationModelOptions: Options for configuring Google Gemini media generation.
GoogleProvider: Provider for Google Gemini native API.
ImageGenerationConfig: Configuration for the image_generation server-side tool.
LanguageModelResult<TOutput extends Object>: Result returned by the model.
LanguageModelUsage: Usage stats for the generation.
LinkPart: A link part referencing external content.
LoggingOptions: Configuration options for logging in the dartantic_ai package.
McpClient: Configuration for connecting to an MCP (Model Context Protocol) server.
MediaGenerationModel<TOptions extends MediaGenerationModelOptions>: Base class for media generation models.
MediaGenerationModelOptions: Base class for media generation model options.
MediaGenerationResult: Streaming chunk returned by a media generation model.
MistralChatModel: Wrapper around Mistral AI Chat Completions API.
MistralChatModelOptions: Options to pass into MistralAI.
MistralEmbeddingsModel: Mistral AI embeddings model implementation.
MistralEmbeddingsModelOptions: Options for Mistral embeddings models.
MistralProvider: Provider for Mistral AI (OpenAI-compatible).
ModelInfo: Model metadata for provider model listing.
ModelStringParser: Parses a model string into a provider name, chat model name, and embeddings model name.
OllamaChatModel: Wrapper around Ollama Chat API that enables to interact with the LLMs in a chat-like fashion.
OllamaChatOptions: Options to pass into Ollama.
OllamaProvider: Provider for native Ollama API (local, not OpenAI-compatible).
OpenAIChatModel: Wrapper around OpenAI Chat API.
OpenAIChatOptions: Generation options to pass into the Chat Model.
OpenAIEmbeddingsModel: OpenAI embeddings model implementation.
OpenAIEmbeddingsModelOptions: OpenAI-specific embeddings model options.
OpenAIProvider: Provider for OpenAI-compatible APIs (OpenAI, Cohere, Together, etc.). Handles API key, base URL, and model configuration.
OpenAIRequestParameters: Request-level parameters for the OpenAI Responses API.
OpenAIResponsesChatModel: Chat model backed by the OpenAI Responses API.
OpenAIResponsesChatModelOptions: Options for configuring the OpenAI Responses chat model.
OpenAIResponsesEventMapper: Maps OpenAI Responses streaming events into dartantic chat results.
OpenAIResponsesHistorySegment: Represents the mapped segment of history sent to the Responses API.
OpenAIResponsesInvocation: Represents a fully-constructed OpenAI Responses API invocation.
OpenAIResponsesInvocationBuilder: Builds OpenAI Responses API invocations from Dartantic messages and options.
OpenAIResponsesMediaGenerationModel: Media generation model built on top of the OpenAI Responses API.
OpenAIResponsesMediaGenerationModelOptions: Options for configuring OpenAI Responses media generation runs.
OpenAIResponsesMessageMapper: Converts between dartantic chat messages and OpenAI Responses API payloads.
OpenAIResponsesOptionsMapper: Utilities for mapping Dartantic options to OpenAI Responses API types.
OpenAIResponsesProvider: Provider for the OpenAI Responses API.
OpenAIServerSideToolContext: Server-side tool configuration context.
Part: Base class for message content parts.
Provider<TChatOptions extends ChatModelOptions, TEmbeddingsOptions extends EmbeddingsModelOptions, TMediaOptions extends MediaGenerationModelOptions>: Provides a unified interface for accessing all major LLM, chat, and embedding providers in dartantic_ai.
StreamingIterationResult: Result from a single streaming iteration
StreamingOrchestrator: Orchestrates the streaming process, coordinating between model calls, tool execution, and message accumulation
TextPart: A text part of a message.
ThinkingConfig: Configuration for enabling Claude's extended thinking. When enabled, responses include thinking content blocks showing Claude's thinking process before the final answer. Requires a minimum budget of 1,024 tokens and counts towards your max_tokens limit.
Tool<TInput extends Object>: A tool that can be called by the LLM.
ToolPart: A tool interaction part of a message.
WebSearchConfig: Configuration for the OpenAI Responses web_search tool.
WebSearchLocation: Approximate geographic hints for web search personalisation.

Enums

AnthropicServerSideTool: Enumerates Anthropic server-side tools with convenience helpers.
AnthropicToolChoiceType: Modes for Anthropic server-side tool selection.
ChatGoogleGenerativeAISafetySettingCategory: Safety settings categorizes.
ChatGoogleGenerativeAISafetySettingThreshold: Controls the probability threshold at which harm is blocked.
ChatMessageRole: The role of a message author.
ChatOpenAIServiceTier: Specifies the latency tier to use for processing the request. This is relevant for customers subscribed to the scale tier service.
FinishReason: The reason the model stopped generating tokens.
GoogleFunctionCallingMode: Controls how the model decides when to call functions.
GoogleServerSideTool: Server-side tools available for Google (Gemini) models.
ImageGenerationQuality: Quality levels for image generation.
ImageGenerationSize: Size options for generated images.
McpServerKind: Represents the type of MCP server connection.
ModelKind: The kind of model supported by a provider.
OpenAIReasoningEffort: Reasoning effort levels for OpenAI Responses models that support thinking.
OpenAIReasoningSummary: Reasoning summary verbosity preference for OpenAI Responses.
OpenAIServerSideTool: OpenAI-provided server-side tools that can be enabled for a Responses call.
ThinkingConfigEnabledType: The type of thinking configuration.
ToolPartKind: The kind of tool interaction.
WebSearchContextSize: Controls how much context is gathered during server-side web search.

Extensions

AnthropicServerSideToolX on AnthropicServerSideTool: Convenience extensions for AnthropicServerSideTool.
MessagePartHelpers on Iterable<Part>: Static helper methods for extracting specific types of parts from a list.

Functions

betaFeaturesForAnthropicTools({List<AnthropicServerToolConfig>? manualConfigs, Set<AnthropicServerSideTool>? serverSideTools}) → List<String>: Computes required beta headers for a set of tools.
mapGoogleMediaFinishReason(Candidate_FinishReason? reason) → FinishReason: Maps Google finish reasons to Dartantic finish reasons.
mapGoogleModalities(List<String>? modalities) → List<GenerationConfig_Modality>: Validates and maps response modalities to Google enums.
mergeAnthropicServerToolConfigs({List<AnthropicServerToolConfig>? manualConfigs, Set<AnthropicServerSideTool>? serverSideTools}) → List<AnthropicServerToolConfig>: Merges explicit tool configs and shorthand server-side tools.
resolveGoogleMediaMimeType(List<String> requested, String? overrideMime) → String: Resolves the best MIME type for Google media generation.

Typedefs

CohereChatModel = OpenAIChatModel: Cohere OpenAI-compatible model.
CohereChatOptions = OpenAIChatOptions: Cohere OpenAI-compatible options.
ContainerFileLoader = Future<ContainerFileData> Function(String containerId, String fileId): Loads a container file by identifier and returns its resolved data.

Exceptions / Errors

OpenAIRefusalException: Exception thrown when OpenAI Structured Outputs API returns a refusal.

dartantic_ai library

Classes

Enums

Extensions

Functions

Typedefs

Exceptions / Errors

dartantic_ai package

dartantic_ai library