Before you dive in
What you’ll find in the original
- The listed price is $0.042 per million input tokens with free output and no published free tier.
- Across fifteen cited runs, the median reported cost is $0.000068 per decision, but workloads vary widely.
- Batching questions reuses state, so adding independent judgments is usually cheaper than resending the same item in separate calls.
Worth knowing
The workload figures are author-reported and heterogeneous; the median is descriptive, not a quote or a guarantee for a new workload.