Skip to content

tinker.types.SamplingPrefillBillingEvent

class tinker.types.SamplingPrefillBillingEvent(StrictBase)

Prompt (prefill) tokens processed while sampling.

Fields:

  • type (Literal['sampling_prefill'])
  • cached (bool) – True when the tokens were served from the prefill cache (billed at the discounted cached rate); False for full-rate prefill
  • token_count (int) – Prefill token count for the bucket