Crate foundry_local_sdk

Source

Expand description

Foundry Local Rust SDK

Local AI model inference powered by the Foundry Local Core engine.

Re-exports§

pub use crate::openai::AudioTranscriptionResponse;
pub use crate::openai::AudioTranscriptionStream;
pub use crate::openai::ChatCompletionStream;
pub use crate::openai::ContentPart;
pub use crate::openai::CoreErrorResponse;
pub use crate::openai::LiveAudioTranscriptionOptions;
pub use crate::openai::LiveAudioTranscriptionResponse;
pub use crate::openai::LiveAudioTranscriptionSession;Deprecated
pub use crate::openai::LiveAudioTranscriptionStream;
pub use crate::openai::TranscriptionSegment;
pub use crate::openai::TranscriptionWord;

Modules§

openai

Structs§

Audio: An audio payload, inline or by URI, with format and PCM layout hints.
AudioSession: An audio-oriented Session for speech tasks (e.g. transcription).
Catalog: The model catalog provides discovery and lookup for all available models.
ChatChoice
ChatChoiceStream
ChatCompletionMessageToolCall
ChatCompletionMessageToolCallChunk
ChatCompletionNamedToolChoice: Specifies a tool the model should use. Use to force the model to call a specific function.
ChatCompletionRequestAssistantMessage
ChatCompletionRequestSystemMessage
ChatCompletionRequestToolMessage: Tool message
ChatCompletionRequestUserMessage
ChatCompletionResponseMessage: A chat completion message generated by the model.
ChatCompletionStreamResponseDelta: A chat completion delta generated by streamed model responses.
ChatSession: A chat-oriented Session with tool registration and turn management.
CompletionUsage: Usage statistics for the completion request.
CreateChatCompletionResponse: Represents a chat completion response returned by model, based on the provided input.
CreateChatCompletionStreamResponse: Represents a streamed chunk of a chat completion response returned by the model, based on the provided input. Learn more.
DownloadBuilder: Builder for configuring and running a model download.
EmbeddingsSession: An embeddings-oriented Session producing dense vectors for text input.
EpDownloadBuilder: Builder for configuring and running execution provider downloads.
EpDownloadResult: Result of a download-and-register execution-provider operation.
EpInfo: Information about an available execution provider.
FoundryLocalConfig: User-facing configuration for initializing the Foundry Local SDK.
FoundryLocalManager: Primary entry point for interacting with Foundry Local.
FunctionCall: The name and arguments of a function that should be called, as generated by the model.
FunctionCallStream
FunctionObject
Image: An image payload, inline or by URI, with an optional format hint (e.g. png).
ItemQueue: A thread-safe, multi-producer / multi-consumer queue of Items.
ItemStream: An asynchronous stream of output Items from Session::process_streaming_request.
Message: A chat message: an author role plus ordered content parts.
Model: The public model type.
ModelInfo: Full metadata for a model variant as returned by the catalog.
ModelSettings: Model-level settings containing a list of parameters.
Parameter: A single parameter key-value pair used in model settings.
PromptTemplate: Prompt template describing how messages are formatted for the model.
Request: A unit of work submitted to a Session.
RequestOptions: Options applied to a Request (or, via Session::set_options, to every request on a session).
Response: The result of processing a Request: output items plus the finish reason and token usage.
Runtime: Runtime information for a model (device type and execution provider).
SearchOptions: Sampling / decoding parameters applied to a request or session.
Session: A stateful inference session bound to a Model.
SpeechResult: A complete speech-recognition result (output-only).
SpeechSegment: One speech-recognition segment (output-only).
SpeechWord: One recognized word within a SpeechSegment (output-only).
Tensor: A numeric tensor: an element data_type, a shape, and the raw little-endian element bytes.
ToolCall: A model-issued request to invoke a named tool/function.
ToolDefinition: A tool the model may call, registered on a ChatSession.
ToolResult: The result of executing a ToolCall, fed back to the model.
Usage: Token accounting for a Response.

Enums§

ChatCompletionMessageToolCalls
ChatCompletionRequestMessage
ChatCompletionToolChoiceOption: Controls which (if any) tool is called by the model. none means the model will not call any tool and instead generates a message. auto means the model can pick between generating a message or calling one or more tools. required means the model must call one or more tools. Specifying a particular tool via {"type": "function", "function": {"name": "my_function"}} forces the model to call that tool.
ChatCompletionTools
ChatFinishReason
ChatResponseFormat: Desired response format for chat completions.
ChatToolChoice: Tool choice configuration for chat completions.
DeviceType: Hardware device type for model execution.
FinishReason: Why generation stopped for a Response.
FoundryLocalError: Errors that can occur when using the Foundry Local SDK.
Item: A single unit of input or output data exchanged with a Session.
ItemType: The kind of payload carried by an Item.
LogLevel: Log level for the Foundry Local service.
MediaSource: The source of an Image or Audio payload: inline bytes or a URI.
MessageRole: The author role of a chat Message.
NativeErrorCode: Stable error codes reported by the native Foundry Local library.
SpeechSegmentKind: Whether a SpeechSegment is an interim hypothesis or a stable result.
TensorDataType: The element data type of a Tensor, mirroring ONNX tensor element types.
TextKind: The subtype of a Item::Text payload.
ToolChoice: How a tool-enabled request should select among available tools.

Traits§

Logger: Application-level logger that the SDK can use to emit diagnostic messages.

Crate foundry_local_sdk

Crate foundry_local_sdk Copy item path

Re-exports§

Modules§

Structs§

Enums§

Traits§

Crate foundry_local_sdk