Interface AIMessageAIOptions

AI evaluator and execution limits. Tokens are weighted AI allowance units, shared across all candidates and expectations in a request. Confidence is model-reported, not a calibrated probability.

Export

Interface

AIMessageAIOptions

Hierarchy

  • AIMessageAIOptions

Properties

maxCandidates maxOutputTokens maxTokens minimumConfidence? model

Properties

maxCandidates

maxCandidates: number

Memberof

AIMessageAIOptions

maxOutputTokens

maxOutputTokens: number

Memberof

AIMessageAIOptions

maxTokens

maxTokens: number

Memberof

AIMessageAIOptions

Optional minimumConfidence

minimumConfidence?: number

Optional advanced abstention threshold for model-reported confidence. Below it the decision is INCONCLUSIVE. It does not guarantee accuracy or cause reevaluation.

Memberof

AIMessageAIOptions

model

model: AIMessageAIOptionsModelEnum

Memberof

AIMessageAIOptions

Settings

Member Visibility

Theme