# ModelSample A general-purpose language model evaluation sample. ``` struct ModelSample where ExpectedValue : Decodable, ExpectedValue : Encodable, ExpectedValue : Sendable ``` ## Overview Accepts string-based prompts and instructions. For multimodal prompts, create a custom [`ModelSampleProtocol`](/documentation/Evaluations/ModelSampleProtocol) conformance or use the [`init(input:expected:expectations:)`](/documentation/Evaluations/ModelSample/init(input:expected:expectations:)) initializer with a prebuilt [`ModelSampleInput`](/documentation/Evaluations/ModelSampleInput). ```swift let sample = ModelSample(prompt: "The capital of France is...", expected: "Paris.") ``` ## Topics ### Creating a sample [`init(prompt:expected:instructions:generationSchema:expectations:)`](/documentation/Evaluations/ModelSample/init(prompt:expected:instructions:generationSchema:expectations:)-7daed) Creates a model sample with string-based prompt and instructions. [`init(prompt:expected:instructions:generationSchema:expectations:)`](/documentation/Evaluations/ModelSample/init(prompt:expected:instructions:generationSchema:expectations:)-8mni) Creates a model sample with a FoundationModels prompt. [`init(input:expected:expectations:)`](/documentation/Evaluations/ModelSample/init(input:expected:expectations:)) Creates a model sample with a prebuilt input. ### Specifying the query [`prompt`](/documentation/Evaluations/ModelSample/prompt) The user’s prompt for this sample. [`promptDescription`](/documentation/Evaluations/ModelSample/promptDescription) A text representation of the prompt, synthesized from its segments. [`instructions`](/documentation/Evaluations/ModelSample/instructions) Optional instructions providing context to the model for this sample. [`instructionsDescription`](/documentation/Evaluations/ModelSample/instructionsDescription) A text representation of the instructions, synthesized from their segments. [`input`](/documentation/Evaluations/ModelSample/input) The bundled language model input (prompt, instructions, schema). ### Specifying expected output [`expected`](/documentation/Evaluations/ModelSample/expected) The expected output for comparison. [`output`](/documentation/Evaluations/ModelSample/output) The expected output value and evaluation expectations. ### Specifying tool expectations [`expectations`](/documentation/Evaluations/ModelSample/expectations) The expected pattern of tool calls for this sample. ### Configuring generation [`generationSchema`](/documentation/Evaluations/ModelSample/generationSchema) The output schema for the model’s response. ### Bundled input and output [`ModelSampleInput`](/documentation/Evaluations/ModelSampleInput) The data a language model receives for evaluation. [`ModelSampleOutput`](/documentation/Evaluations/ModelSampleOutput) The expected output value and evaluation expectations for a sample. ### Protocols [`ModelSampleProtocol`](/documentation/Evaluations/ModelSampleProtocol) A type that defines language model evaluation samples with prompt, instructions, and expectations. [`SampleProtocol`](/documentation/Evaluations/SampleProtocol) A type that defines evaluation samples. ## Relationships ### Conforms To [`ModelSampleProtocol`](/documentation/Evaluations/ModelSampleProtocol) [`Escapable`](/documentation/Swift/Escapable) [`SendableMetatype`](/documentation/Swift/SendableMetatype) [`Sendable`](/documentation/Swift/Sendable) [`Decodable`](/documentation/Swift/Decodable) [`SampleProtocol`](/documentation/Evaluations/SampleProtocol) [`Copyable`](/documentation/Swift/Copyable) [`Encodable`](/documentation/Swift/Encodable) --- Copyright © 2026 Apple Inc. All rights reserved. | [Terms of Use](https://www.apple.com/legal/internet-services/terms/site.html) | [Privacy Policy](https://www.apple.com/privacy/privacy-policy)