AxAIOpenAIBaseArgs
Defined in: https://github.com/ax-llm/ax/blob/e690b4823b8309e258594064abe941d4cff6ef89/src/ax/ai/openai/api.ts#L281
Type Parameters
Properties
| Property | Type | Description |
|---|
apiKey? | string | - |
apiURL? | string | - |
chatReqUpdater? | ChatReqUpdater<TModel, TChatReq> | - |
chatRespProcessor? | ChatRespProcessor | - |
chatStreamRespProcessor? | ChatStreamRespProcessor | - |
config | Readonly<AxAIOpenAIConfig<TModel, TEmbedModel>> | - |
credentialProfile? | string | - |
credentialProvider? | AxAICredentialProvider | - |
modelInfo | readonly AxModelInfo[] | - |
models? | AxAIInputModelList<TModel, TEmbedModel, TModelKey> | - |
options? | Readonly<AxAIServiceOptions & object> | - |
promptCacheBreakpoints? | boolean | Whether the compatible API accepts explicit per-content cache breakpoints. |
promptCaching? | boolean | Opt in to OpenAI prompt caching on GPT-5.6+ models. Off by default because this request builder is shared: Azure OpenAI is typed on the same model enum, so a gpt-5.6-* deployment would otherwise pick up parameters its API version may reject. Official OpenAI and explicitly compatible named profiles opt in deliberately. |
promptCachingAllModels? | boolean | Apply the compatible prompt-cache wire format without OpenAI model-family gating. |
realtime? | RealtimeAdapter<TModel> | - |
supportFor | | AxAIFeatures | (model: TModel) => AxAIFeatures | - |