Table of Contents

Class GenerativeRequest<TSelf, TInput, TOutput, TChunk, TEvent>

Namespace
Glitch9.AIDevKit

Abstract base class for all generative AI tasks. Provides common properties and methods for handling prompts, models, outputs, and execution.

public abstract class GenerativeRequest<TSelf, TInput, TOutput, TChunk, TEvent> : FluentApiRequest<TSelf, TOutput>, IGenerativeRequest, ISequentialRequest where TSelf : GenerativeRequest<TSelf, TInput, TOutput, TChunk, TEvent> where TInput : IPrompt where TOutput : IGeneratedOutput where TEvent : IGenerativeEvent<TChunk, TOutput>

Type Parameters

TSelf
TInput
TOutput
TChunk
TEvent
Inheritance
object
FluentApiRequest<TSelf, TOutput>
GenerativeRequest<TSelf, TInput, TOutput, TChunk, TEvent>
Implements
Derived
Inherited Members
Extension Methods

Constructors

GenerativeRequest() 

protected GenerativeRequest()

GenerativeRequest(TInput) 

protected GenerativeRequest(TInput prompt)

Parameters

prompt TInput

Properties

Api

The cloud service provider associated with this task. Automatically resolved from the selected model if not explicitly set.

public override Api Api { get; set; }

Property Value

Api

ArchiveHistory

If true, the prompt and output history will be archived (hidden) after saving.

public bool ArchiveHistory { get; set; }

Property Value

bool

MaxN

The maximum number of outputs allowed for the selected model.

Added new on 2025.07.17

public int MaxN { get; }

Property Value

int

MeasureLatency

If true, the system will measure the elapsed time from when the request is sent until the response is received.

public bool MeasureLatency { get; set; }

Property Value

bool

Model

The identifier of the model to use for this task. Matches the model deployment name or model ID provided by the API.

public string Model { get; set; }

Property Value

string

ModelType 

public abstract ModelType ModelType { get; }

Property Value

ModelType

N

The number of responses to generate. Must be between 1 and 10.

public NCount N { get; set; }

Property Value

NCount

OnLatencyMeasured 

public Action<double> OnLatencyMeasured { get; set; }

Property Value

Action<double>

OnRecordCreated

The event listener that will be called when a new prompt record is created for this request.

public Action<AIRequestRecord> OnRecordCreated { get; set; }

Property Value

Action<AIRequestRecord>

OutputFileNote 

public string OutputFileNote { get; set; }

Property Value

string

Prompt

The prompt used for generation. Can be a text, audio, or custom prompt type.

public virtual TInput Prompt { get; set; }

Property Value

TInput

RecordHistory

Whether prompt history should be recorded for this task. Default is controlled by SavePromptHistoryOnRuntime.

public bool RecordHistory { get; set; }

Property Value

bool

RecordId

A fixed seed value to make generation deterministic and repeatable.

public string RecordId { get; set; }

Property Value

string

RecordMergeOptions

The record merge settings for this request.

  • Added new on 2025.11.27
public RecordMergeOptions RecordMergeOptions { get; set; }

Property Value

RecordMergeOptions

RequestedAt

The timestamp when the request was initiated. Only set when CountdownTimer is true. 

public DateTime RequestedAt { get; }

Property Value

DateTime

SaveOutputs

Whether generated outputs should be saved to disk.

public bool SaveOutputs { get; set; }

Property Value

bool

Seed

Random seed for deterministic sampling (when supported):

  • Purpose – Reproduce the same output across runs with identical inputs.
  • Scope – Holds only if provider, model/deployment, version, and all params are unchanged.
  • null – Lets the service choose a random seed (non-deterministic).
  • Range – 0–4,294,967,295 (32-bit).
  • Support – Some models/services ignore seeds; if unsupported, this has no effect.
public Seed Seed { get; set; }

Property Value

Seed

Methods

CreateStreamAsync() 

public UniTask<IRESTStream<TEvent>> CreateStreamAsync()

Returns

UniTask<IRESTStream<TEvent>>

CreateStreamAsyncInternal() 

protected abstract UniTask<IRESTStream<TEvent>> CreateStreamAsyncInternal()

Returns

UniTask<IRESTStream<TEvent>>

ExecuteAsync(CancellationToken)

Executes the task and returns the output of type TOutput. This is the main entry point for executing the task.

public override sealed UniTask<TOutput> ExecuteAsync(CancellationToken ct = default)

Parameters

ct CancellationToken

Returns

UniTask<TOutput>

ExecuteAsyncT<T>(CancellationToken) 

public UniTask<T> ExecuteAsyncT<T>(CancellationToken ct = default) where T : IGeneratedOutput

Parameters

ct CancellationToken

Returns

UniTask<T>

Type Parameters

T

GetPromptToRecord() 

public IPrompt GetPromptToRecord()

Returns

IPrompt

GetRequestParameters() 

public virtual Dictionary<string, object> GetRequestParameters()

Returns

Dictionary<string, object>

ProcessPostRequestTasks(IGeneratedOutput) 

public void ProcessPostRequestTasks(IGeneratedOutput output)

Parameters

output IGeneratedOutput

ResolveOutputPath(string) 

protected override sealed string ResolveOutputPath(string rawOutputPath = null)

Parameters

rawOutputPath string

Returns

string

SetCount(int)

Sets the number of outputs to generate (e.g., number of images or responses).

public TSelf SetCount(int count)

Parameters

count int

Returns

TSelf

SetGeneratorWindowSettings(string, string, bool, Action<double>) 

public void SetGeneratorWindowSettings(string sender, string outputPath, bool measureLatency = true, Action<double> onLatencyMeasured = null)

Parameters

sender string
outputPath string
measureLatency bool
onLatencyMeasured Action<double>

SetMeasureLatency(Action<double>) 

public TSelf SetMeasureLatency(Action<double> onLatencyMeasured)

Parameters

onLatencyMeasured Action<double>

Returns

TSelf

SetMeasureLatency(bool, Action<double>) 

public TSelf SetMeasureLatency(bool measureLatency = true, Action<double> onLatencyMeasured = null)

Parameters

measureLatency bool
onLatencyMeasured Action<double>

Returns

TSelf

SetModel(Api, string)

Sets the AI model with Provider API and model ID.

  • Added new on 2025.07.17
public TSelf SetModel(Api api, string modelId)

Parameters

api Api
modelId string

Returns

TSelf

SetModel(Model)

Sets the AI model to be used for generation.

public TSelf SetModel(Model model)

Parameters

model Model

Returns

TSelf

SetOutputPath(string, string)

Sets the download path for the generated content(s). This is the path where the generated content(s) will be saved.

public TSelf SetOutputPath(string outputPath = null, string outputNote = null)

Parameters

outputPath string

The path where the generated content(s) will be saved.

outputNote string

Returns

TSelf

SetRecordHistoryOptions(bool?, bool?, string, TInput)

Enables or disables prompt history for this task.

public TSelf SetRecordHistoryOptions(bool? record = null, bool? archive = null, string recordId = null, TInput promptToRecord = default)

Parameters

record bool?
archive bool?
recordId string
promptToRecord TInput

Returns

TSelf

SetRecordMergeOptions(AIRequestRecord, Func<AIRequestRecord, AIRequestRecord, AIRequestRecord>) 

public TSelf SetRecordMergeOptions(AIRequestRecord recordToMerge, Func<AIRequestRecord, AIRequestRecord, AIRequestRecord> mergeFunction)

Parameters

recordToMerge AIRequestRecord
mergeFunction Func<AIRequestRecord, AIRequestRecord, AIRequestRecord>

Returns

TSelf

SetRecordMergeOptions(RecordMergeOptions) 

public TSelf SetRecordMergeOptions(RecordMergeOptions recordMergeOptions)

Parameters

recordMergeOptions RecordMergeOptions

Returns

TSelf

SetSeed(long)

Sets a fixed seed to make generation deterministic and repeatable.

public TSelf SetSeed(long seed)

Parameters

seed long

Returns

TSelf