The parsed, Zod-validated object matching the provided schema.
The raw LLM output text before parsing.
Aggregated token usage across all attempts (including retries).
Reason the model stopped generating on the final successful attempt. Mirrors the finish reasons from generateText.
Provider identifier used for the run.
Resolved model identifier used for the run.
Optional fallbackProvider-fallback trail aggregated across all attempts. fired is true
when any attempt used or tried a non-primary provider — including a
fallback on a failed attempt that a later attempt recovered from on the
primary. Callers use this to flag a degraded run.
The completed result returned by generateObject.