aiapicostcalcUnit economicsPricesBrandsCompareBenchmarksSkills & PluginsFind my modelGuidesChanges

Cost calculator vs OpenRouter vs provider pricing pages

Short answer: A provider pricing page is the authoritative rate card, a calculator turns those rates into cost per call and per user, and a marketplace such as OpenRouter passes provider prices through but charges 5.5% when you buy credits on Standard (8% on Business). Use the provider page as the source of truth and compare the marketplace's total, fee included, against the same token counts.

Last verified 2026-11-18 · Sources: official provider pricing pages, copied into src/prices.json (checked 2026-11-18) · Report an error

Three common ways to work out what an LLM API will cost: a calculator like this one, the price list on a model marketplace such as OpenRouter, or the provider's own pricing page. They answer different questions. This page compares them neutrally; each fact about another service comes from that service's own page, linked below with the date we read it.

Side by side

aiapicostcalcOpenRouter models listProvider pricing page
What it isFree browser calculator: cost per call, per month, per user and minimum price at a target marginModel marketplace and unified API; its models page lists per-token prices for many providersOfficial rate card for one provider's models
Price sourceCopied by hand from each provider's official page; source link and check date on every modelProvider prices passed through without inference markup (per its FAQ)Authoritative for that provider
FeesNone; the site does not resell API access5.5% on credit purchases (Standard), 8% (Business); no markup on inference (read 2026-11-09)Billed by the provider; batch and caching discounts listed there
Cross-provider ranking for your workloadYes, every tracked model ranked for the token counts you enterLists and sorts models by price; no per-user budget mathsNo, one provider only
Caching and batch detailCached-input rate per model; batch covered in a guideVaries by model listingMost complete (e.g. Anthropic lists 5-min and 1-hour cache-write multipliers)
Needs an accountNoAccount and credits to call modelsAccount to call models

Which one to use

Citable example: one RAG call on Claude Sonnet 5.5

WorkloadCost per callPer 1,000 calls
20,000 input / 1,000 output tokens, no cache$0.0500$50.00
Same, 80% of input served from cache$0.0196$19.60
Rates: $2 input, $0.1 cached input, $10 output per 1M tokens, from the official pricing page (checked 2026-11-15). Cache-write charges excluded. Calculated 2026-10-27.

Sources (read 2026-10-27)

Questions

Is OpenRouter more expensive than calling the provider directly?

OpenRouter's FAQ says it passes through the underlying providers' inference prices without markup and charges a fee when you purchase credits. Check its pricing page for the current credit fee before comparing totals.

Why use a calculator if the provider already publishes prices?

A pricing page lists rates per 1M tokens. A calculator turns those rates into cost per call, per month and per user for your token counts, and ranks models for the same workload.

Which source should I trust for billing?

Your provider's own pricing page and invoice. aiapicostcalc copies those rates by hand with a check date on every row; use it to estimate and compare, then confirm with the provider.

Does OpenRouter charge a fee on top of provider prices?

OpenRouter passes provider inference prices through without a markup and charges a fee when you buy credits: 5.5% on Standard and 8% on Business, never per request, per its pricing page (read 2026-11-09). Add that credit fee to the pass-through prices before comparing with direct billing; the /openrouter-pricing page does that arithmetic.

Next step → Estimate your own monthly bill

Price data last checked 2026-11-18 against official sources · how we verify

Price-change alerts

One short email when a provider changes a price or a promo ends. No spam, unsubscribe any time.