About TokenRate

Model catalog built 2026-10-21

TokenRate is a free, open token-and-pricing calculator for developers, product teams, and finance owners who run workloads on AI APIs. We answer two questions before you ship: how much will this prompt cost? and how many words can my budget buy across models?

Who runs TokenRate

TokenRate is built and maintained by Elliott Crosby (GitHub). The site combines model price data with calculators and practical guides so readers can estimate the cost of their own workloads. Rates and billing rules change, so each guide links to provider documentation and the calculators show their limitations. If you find an error, send a correction.

Coverage

As of 2026-10-21, TokenRate tracks 304 AI models across 29 providers (Anthropic, OpenAI, Google, Meta, DeepSeek, xAI, Mistral), publishes 17 head-to-head comparisons, and maintains 17 reference guides on tokenization and pricing. Every model has a dedicated page with input/output prices, context window, output limit, strengths, weaknesses, and cost examples.

Methodology

Live pricing for the calculator is pulled from the OpenRouter API, which publishes model and provider-route rates. Some calculator pages use a catalog generated at build time; pages with a live price lookup cache that lookup for up to an hour. The catalog build date does not prove that every provider SKU was individually verified that day.

Individual model pages may combine catalog prices with a live OpenRouter lookup. A Live badge identifies a successful lookup; otherwise a reference price is shown. The displayed date is the catalog update date. Aggregator route prices can differ from a direct provider invoice.

For mission-critical decisions, always confirm with the provider’s official pricing page ( Anthropic, OpenAI, Google ). Check the exact model, service tier and billing category on your invoice.

Token estimation

For English text, TokenRate uses the standard 4 characters per token approximation for a quick planning estimate. Error varies by language, formatting, code, and model tokenizer. For exact counts on critical workloads, use provider tools: tiktoken for OpenAI, the Anthropic SDK for Claude, and Google’s countTokens API for Gemini.

What we do not do

We do not take payment for placement on rankings or comparisons. We do not take commissions on model purchases or API signups. Rankings are derived from public pricing and benchmarks, and verdicts on comparison pages reflect our reading of published evaluations — not a vendor relationship.

Is TokenRate free?

Yes — no account, no sign-up, no credit card. The site may display advertising after approval; the calculators and guides remain free. See our privacy policy for details on how advertising cookies work.

Contact & corrections

Pricing corrections, missing models, or feedback? Get in touch → or email nodea.ai@gmail.com. Include the model ID and a link to the provider’s price page when reporting a rate.