TokenMeter is a single-purpose tool: it tells you what a large language model API call will actually cost you before you ship it, not after the invoice arrives.
Almost every cost calculator on the web multiplies your token count by the headline price. That number is wrong for most real workloads, and wrong in a consistent direction — it overstates your bill.
Three mechanisms account for the difference, and all three are billed as separate line items that naive calculators ignore:
TokenMeter applies all three automatically and ranks every model by resulting monthly cost, so the question answered is "which model is cheapest for my workload", not "which model has the smallest sticker price".
Every rate in the table is transcribed from the provider's own published pricing page and carries a verification date. The current snapshot is dated September 2026 and covers 25 models across OpenAI, Anthropic, Google, xAI and DeepSeek. Several rates are promotional and time-limited — the table marks these explicitly, because a price that silently doubles on a published date is worse than no price at all.
LLM pricing moves fast, often monthly. Treat this as a dated snapshot rather than a live feed, and confirm against the official page before you commit to a contract or publish a customer-facing price. Every figure links to its source at the foot of the calculator.
Errors in a cost tool are expensive, so the process is deliberate:
If you find an error, tell us on the contact page with the model and the rate. We will verify against the provider and correct it.
No accounts, no tracking of what you type, no API key required, and no data leaves your browser. See the privacy policy.
TokenMeter is not affiliated with, endorsed by, or sponsored by OpenAI, Anthropic, Google, xAI, DeepSeek or any other model provider. Model names are trademarks of their respective owners. We take no position on which model you should use — only on what it costs.