· CostMyTokens · Guides · 4 min read
How CostMyTokens Verifies LLM API Prices (Methodology)
How CostMyTokens verifies LLM API prices: official pages plus a second source, per-model verification dates, weekly bulk re-checks, and why we never invent unverified gateway rates.
Quick answer: CostMyTokens is a free LLM API cost calculator that uses official provider prices, verifies them weekly, and never mixes in unverified gateway rates. Every rate comes from the maker’s pricing page and a second independent source; each model shows a verification date; the full dataset is re-checked on a weekly cycle; new models land within about 48 hours once official prices are confirmed. Unknown fields stay Not published — we do not invent them. Calculations run in your browser; pasted text is never uploaded.
Disclosure: some links elsewhere on this site to Novita AI are affiliate links — we may earn a commission at no extra cost to you. Affiliate status never changes a listed price, a calculator result, or ranking order (rankings are by price only). Read our disclosure.
How we verify each price
- Official page first. We copy list prices from the provider’s own pricing documentation (OpenAI, Anthropic, Google, DeepSeek, xAI, and others we track).
- Second source. Before publish, we cross-check against a second, independent public source (another official page, docs mirror, or vendor changelog) so a single stale scrape cannot silently ship.
- Per-model verification date. Every model page and dataset row carries the date that rate was last checked, so you can see how fresh that SKU is — not only a site-wide stamp.
- Weekly bulk pass. The full dataset is re-checked on roughly a weekly cycle. The About page surfaces the latest full-dataset (bulk) verification pass date from our pricing data file.
- New models ~48 hours. Once a maker publishes confirmable official prices, we aim to add the model within about 48 hours.
- Future changes noted early. When a provider announces a dated price change (for example Gemini 4 Argon’s scheduled rate change), we note it on the model page in advance and apply it on the effective day.
What we refuse to invent
- Not published fields. If a provider has not published a figure (cached input, batch, etc.), we show Not published instead of estimating. An empty cell is honest; an invented number is not.
- Unverified gateway rates. The calculator defaults to official list prices. Gateway rows appear only when we can verify a public host price. If there is no public lower gateway rate for a SKU, we do not invent a cheaper line.
- No accuracy %, no client savings stories. We do not publish an unverifiable “accuracy rate,” customer case studies, or “most accurate calculator” claims. Trust the methodology and the dated sources — not marketing adjectives.
- Closed models ≠ gateway discounts. A gateway hosting an open-weight SKU is not a discount on GPT / Claude / Gemini. We keep those paths labeled separately; see Official API vs gateway pricing.
How the calculator works
For one API call:
Cost = (input tokens × input price + output tokens × output price + cached tokens × cached price) ÷ 1,000,000
Multiply by call volume for a daily or monthly estimate. You can switch Standard / Batch tiers where a provider publishes a Batch discount.
Token counting is provider-specific:
- OpenAI models use the real o200k tokenizer in the browser.
- Where a provider publishes no official browser tokenizer, the count is labeled an estimate.
All math runs locally in your browser. Pasted prompt/output text is never uploaded. There is no user account and the tool is free to use. Details: About & methodology and Privacy.
Affiliate independence
CostMyTokens is an indie project operated by Kim Han. It is not affiliated with OpenAI, Anthropic, Google, or any gateway. Rankings and comparisons are ordered by price only.
Some outbound links (currently Novita AI) are marked (affiliate). Commissions never change a listed price, a calculator result, or the order of a ranking. Full wording: Affiliate disclosure.
FAQ
How does CostMyTokens verify prices? From the provider’s official pricing page, cross-checked against a second independent source. Each model has a verification date; the full dataset is re-checked about weekly; new models are added within about 48 hours once official prices are confirmed.
Does CostMyTokens mix official and gateway prices? No. The default view is official list prices. Gateway rates appear only when we can verify them from a public source. Unverified gateway rates are never mixed into official rows.
Is CostMyTokens free? Do I need an account? Yes, it is free. There is no user account. Calculations stay in your browser.
Who runs CostMyTokens? Kim Han, as an independent one-person indie project. Not a maker or gateway affiliate.
Does affiliate commission change prices or rankings? No. Affiliate links (currently Novita AI) never change listed prices, calculator results, or sort order — rankings are by price only.
What if a price looks wrong? Email [email protected] or use the Contact page. We correct dated rows when a public source shows a change.