Methodology & pricing sources
Last checked . Prices snapshot on the calculator still shows September 8, 2026 until the next rate-file refresh.
TokenCost multiplies the tokens you enter by the selected model’s published USD per million input and output tokens. Math runs entirely in your browser from a static rate file — we do not call provider billing APIs.
How numbers are derived
- Cost ≈ (input_tokens / 1e6 × input_$per_M) + (output_tokens / 1e6 × output_$per_M).
- Optional daily/monthly projections multiply per-request cost by your request volume.
- Cached-input or long-context repricing is noted per SKU when the vendor documents it; the calculator uses the SKU line you select.
Provider list-price sources
- OpenAI API pricing — GPT family list rates.
- Anthropic pricing — Claude API list rates.
- Google AI / Gemini pricing — Gemini API list rates.
- DeepSeek and other SKUs: vendor public pricing pages linked from the model row on /pricing.
We refresh the rate file when those public pages move enough to change estimates. Editorial last checked: September 19, 2026. The on-page “prices last updated” string tracks the rate-file commit date.
Estimates only. Not a quote, invoice, or financial advice. Enterprise discounts, batch APIs, and regional billing can differ.
Layout scaffolding and how-to prose may be drafted with generative AI; rate tables and cost formulas are taken from documented provider sources and human-reviewed before we publish.