OptionalmaxMaximum number of retries when the generated output fails validation.
OptionalmaxMaximum number of tokens to generate.
The Vercel AI SDK language model to use.
OptionalonOptional callback invoked with each text chunk streamed from the model. When provided, the LLM response is streamed instead of returned all at once.
The incremental text chunk from the model.
OptionaltemperatureSampling temperature. Lower values produce more deterministic output.
Some models (kimi-k2.6 with my test) only accept a temperature of 1.0.
Common options shared by AI resume operations.