Skip to content

AI model cost tools

Both tools run entirely in your browser and price every model with the same source-verified rates used across this site — official list prices only, with tier, cache, and long-context rules applied. No account, no API key, and nothing you paste or upload leaves your device.

Pick the tool that matches your question

One forward-looking, one backward-looking — both answer in the same currency: your monthly bill.

Model cost calculator

Estimate monthly API cost for the GPT-6 family and Claude models: service tiers (batch/flex/fast/ultrafast), cached reads and cache writes, per-request long-context banding, and CSV export. Unpublished combinations show Unsupported, never a guess.

Open the Model cost calculator →

Usage audit (local CSV)

Recompute your spend from your own OpenAI or Anthropic usage export using official rates, reconcile it against your cost file, and model a model switch. Every figure carries an exact / estimate / range label, and the files never leave your browser.

Open the Usage audit (local CSV) →

What both tools share

  • The same pricing rows, hand-verified against vendor pages and dated · pricing data (verified 2026-10-09)
  • Tier multipliers (batch/flex at half price, fast at 2×, ultrafast at 6×), per-request long-context banding, and cache reads/writes billed where the vendor publishes them.
  • Published limits: when a tier or a model/rate combination has no published price, the tool says Unsupported instead of extrapolating.
  • Browser-local processing: token counts, CSVs, and exports stay on your device; the only network request is optional, aggregate product analytics with no PII.

Not sure which model you're pricing?

Start from a decision page instead — each comparison leads with the verdict, then shows the arithmetic.