AgentOptimizationOptions Class

Definition

Namespace:

Azure.AI.Projects.Agents

Assembly:

Azure.AI.Projects.Agents.dll

Package:

Azure.AI.Projects.Agents v3.0.0-beta.1

Source:

AgentOptimizationOptions.cs

Tuning knobs and run-mode for an optimization job.

public class AgentOptimizationOptions : System.ClientModel.Primitives.IJsonModel<Azure.AI.Projects.Agents.AgentOptimizationOptions>, System.ClientModel.Primitives.IPersistableModel<Azure.AI.Projects.Agents.AgentOptimizationOptions>

type AgentOptimizationOptions = class
    interface IJsonModel<AgentOptimizationOptions>
    interface IPersistableModel<AgentOptimizationOptions>

Public Class AgentOptimizationOptions
Implements IJsonModel(Of AgentOptimizationOptions), IPersistableModel(Of AgentOptimizationOptions)

Inheritance

Object

AgentOptimizationOptions

Implements

IJsonModel<AgentOptimizationOptions> IPersistableModel<AgentOptimizationOptions> IPersistableModel<T>

Constructors

Name	Description
AgentOptimizationOptions()	Initializes a new instance of AgentOptimizationOptions.

Properties

Name	Description
EvalModel	Model deployment used for evaluation. Defaults to server config (typically 'gpt-4o').
EvaluationLevel	Evaluation granularity. Null/omitted means per-item single-turn. Set to 'conversation' for per-conversation multi-turn simulation scoring.
MaxCandidates	Maximum number of optimization candidates to generate. Must be >= 1. Default: 5.
MaxStalls	Maximum number of consecutive reflective minibatch rejections before stopping early. A 'stall' occurs when the optimizer proposes a prompt change, evaluates it on a small subset, and the score does not improve — so no full validation-set evaluation is triggered. The counter resets whenever a minibatch passes and its full-validation score beats the current best. Only a sustained plateau of max_stalls consecutive minibatch failures triggers the stop. The service defaults to 5 if a value is not specified by the caller. Must be >= 1 when set.
OptimizationConfig	Per-target-attribute configuration overrides. Contains skills, tools, system_prompt for the agent, plus model space for model optimization. To assign an object to the value of this property use FromObjectAsJson<T>(T, JsonSerializerOptions). To assign an already formatted json string to this property use FromString(String). Examples: BinaryData.FromObjectAsJson("foo"). : Creates a payload of "foo". BinaryData.FromString("\"foo\""). : Creates a payload of "foo". BinaryData.FromObjectAsJson(new { key = "value" }). : Creates a payload of { "key": "value" }. BinaryData.FromString("{\"key\": \"value\"}"). : Creates a payload of { "key": "value" }.
OptimizationModel	Model deployment for optimization reasoning (must be gpt-5 family). Falls back to the default eval model when not set.

Methods

Name	Description
JsonModelCreateCore(Utf8JsonReader, ModelReaderWriterOptions)
JsonModelWriteCore(Utf8JsonWriter, ModelReaderWriterOptions)
PersistableModelCreateCore(BinaryData, ModelReaderWriterOptions)
PersistableModelWriteCore(ModelReaderWriterOptions)

Explicit Interface Implementations

Name	Description
IJsonModel<AgentOptimizationOptions>.Create(Utf8JsonReader, ModelReaderWriterOptions)
IJsonModel<AgentOptimizationOptions>.Write(Utf8JsonWriter, ModelReaderWriterOptions)
IPersistableModel<AgentOptimizationOptions>.Create(BinaryData, ModelReaderWriterOptions)
IPersistableModel<AgentOptimizationOptions>.GetFormatFromOptions(ModelReaderWriterOptions)
IPersistableModel<AgentOptimizationOptions>.Write(ModelReaderWriterOptions)