mojentic-openai/com.mojentic.openai/OpenAIModelInfo

OpenAIModelInfo

data class OpenAIModelInfo(val id: String, val modelType: ModelType, val contextWindow: Int?, val maxOutputTokens: Int?, val supportsTools: Boolean, val supportsStreaming: Boolean, val supportsVision: Boolean, val supportedTemperatures: List<Double>?, val supportsChatApi: Boolean, val supportsCompletionsApi: Boolean, val supportsResponsesApi: Boolean)

Static metadata about an OpenAI model.

Mirrors the ModelCapabilities interface in the other Mojentic ports — a lookup record describing the context window, feature flags, temperature restrictions, and per-API support a broker / chat session might consult before issuing a call. Update conservatively: stale entries are far better than wrong ones.

Constructors

OpenAIModelInfo

constructor(id: String, modelType: ModelType, contextWindow: Int?, maxOutputTokens: Int?, supportsTools: Boolean, supportsStreaming: Boolean, supportsVision: Boolean, supportedTemperatures: List<Double>?, supportsChatApi: Boolean, supportsCompletionsApi: Boolean, supportsResponsesApi: Boolean)

Properties

contextWindow

val contextWindow: Int?

maximum context window in tokens, or null if unknown

val id: String

the model identifier as used by the OpenAI API

maxOutputTokens

val maxOutputTokens: Int?

maximum output tokens per response, or null if unknown

modelType

val modelType: ModelType

classification used to pick token-limit parameters

supportedTemperatures

val supportedTemperatures: List<Double>?

null means all temperatures are supported; an empty list means the temperature parameter is not allowed; a non-empty list enumerates the only permitted values

supportsChatApi

val supportsChatApi: Boolean

whether the model is reachable via the Chat Completions API

supportsCompletionsApi

val supportsCompletionsApi: Boolean

whether the model is reachable via the legacy Completions API

supportsReasoningEffort

val supportsReasoningEffort: Boolean

Whether this model is a reasoning model — equivalent to checking modelType against ModelType.REASONING. Reasoning models use max_completion_tokens instead of max_tokens and accept a reasoning_effort parameter.