Expand description
Review cost transparency.
A small pricing table for the LLM models DiffLore actually calls, plus
a helper to estimate the USD cost of a single review turn from the
provider’s returned usage block.
The table is intentionally conservative: unknown models return None
rather than guessing, so downstream code can persist NULL and the
cloud aggregation skips the row instead of under-reporting.
Structs§
- Model
Pricing - Per-1K-token pricing for a single LLM model. Mirrors the public pricing pages for each provider as of the plan date; update when providers change their rate cards.
Functions§
- estimate_
cost_ usd - Compute the estimated USD cost of a single LLM call.
- pricing_
for - Look up the per-1K pricing for a model. Returns
Nonefor unknown identifiers so callers can recordNULLand avoid under/over-reporting.