Skip to main content

Type Alias: AIBatchResult<T>

AIBatchResult<T> = { cachedInputTokens?: number; error?: undefined; id: string; inputTokens: number; outputTokens: number; prompt: string; reasoningTokens?: number; reportedCostUsd?: number; result: T; servedBy?: string; status: "succeeded"; validated?: boolean; } | { cachedInputTokens?: number; error: string; id: string; inputTokens: number; outputTokens: number; prompt: string; reasoningTokens?: number; reportedCostUsd?: number; responseText?: string; result?: undefined; servedBy?: string; status: "failed"; validated?: boolean; }

Defined in: packages/workflow-engine/src/ai/types.ts:544

Result of a single request in a batch

Type Parameters​

T​

T = string

Union Members​

Type Literal​

{ cachedInputTokens?: number; error?: undefined; id: string; inputTokens: number; outputTokens: number; prompt: string; reasoningTokens?: number; reportedCostUsd?: number; result: T; servedBy?: string; status: "succeeded"; validated?: boolean; }

cachedInputTokens?​

optional cachedInputTokens?: number

Input tokens served from the prompt cache, when reported.

error?​

optional error?: undefined

id​

id: string

The request ID (matches the id from AIBatchRequest)

inputTokens​

inputTokens: number

Input tokens used

outputTokens​

outputTokens: number

Output tokens used

prompt​

prompt: string

Original prompt (may be empty if not available from provider)

reasoningTokens?​

optional reasoningTokens?: number

Reasoning tokens the model emitted, when reported.

reportedCostUsd?​

optional reportedCostUsd?: number

USD cost the transport reported for this request (OpenRouter with usage accounting), when it reports one. Absent on the vendor transports; recordResults then bills the batch estimate.

result​

result: T

The parsed result. When a schema was provided and validation succeeded, validated is true. When no schema was available at retrieval time (the default after suspend/resume), this is the unvalidated parsed JSON or string output, and validated is false.

servedBy?​

optional servedBy?: string

The endpoint that served the request, when known.

status​

status: "succeeded"

validated?​

optional validated?: boolean

True only when a schema was available at retrieval time and the response was successfully parsed and validated against it. False/undefined when no schema was available (e.g. after a suspend/resume without re-supplying schemas).

Type Literal​

{ cachedInputTokens?: number; error: string; id: string; inputTokens: number; outputTokens: number; prompt: string; reasoningTokens?: number; reportedCostUsd?: number; responseText?: string; result?: undefined; servedBy?: string; status: "failed"; validated?: boolean; }

cachedInputTokens?​

optional cachedInputTokens?: number

Input tokens served from the prompt cache, when reported.

error​

error: string

Error message describing why the request failed.

id​

id: string

inputTokens​

inputTokens: number

outputTokens​

outputTokens: number

prompt​

prompt: string

reasoningTokens?​

optional reasoningTokens?: number

Reasoning tokens the model emitted, when reported.

reportedCostUsd?​

optional reportedCostUsd?: number

USD cost the transport reported for this request. A response the provider served but that failed local JSON parsing or schema validation was still billed, so the figure is kept here for recordResults. Absent when the provider itself failed the request.

responseText?​

optional responseText?: string

The model's raw reply when the request reached the model but its output could not be parsed or validated (absent when the provider itself failed the request). Stored on the accounting row.

result?​

optional result?: undefined

No result is available for a failed request.

servedBy?​

optional servedBy?: string

The endpoint that served the request, when known.

status​

status: "failed"

validated?​

optional validated?: boolean

Always false on failed requests.