Know what your prompts cost before you ship them
Paste a prompt and get exact token counts plus a per-model cost breakdown across 44 models. No sign-up needed to start.
Models
3 selectedOpenAI
These 3 are free. Open the full analyzer to see all 44 models across 9 providers.
Free tier: 3 models, 10,000 characters, one input at a time. Pro unlocks all 44 models, batches, cost projection and CSV export.
Built for people who pay the inference bill
TokenHub is a measurement tool. It counts and prices tokens locally and never sends your text to a model provider.
Exact counts, not guesses
OpenAI models are tokenized with the real tiktoken-compatible encodings: o200k_base, cl100k_base and o200k_harmony. No four-characters-per-token hand waving.
See the money before you ship
Every count is priced against published input and output rates, so you can compare a cheap model at high volume against an expensive one at low volume.
Model your prompt cache
Cache hits are billed at a fraction of the normal input rate. Set your expected hit rate and see what caching is actually worth on your workload.
Catch context overruns early
Prompt plus expected completion is checked against each model's context window, so a request that cannot fit is flagged before it fails in production.
Batch a whole prompt library
Paste up to 200 prompts at once and get per-prompt and aggregate totals, then copy the result straight out as CSV or JSON.
Your rates, not the list price
On an enterprise agreement or a reseller discount? Override any model's input and output rate and the whole comparison recalculates.
44
models in the catalogue
16
tokenized with the vendor's real encoding
0
of your text sent to any model provider
Straightforward pricing
$9 per month, or $79 per year. Cancel yourself at any time, and there is a 14-day no-questions-asked refund on every purchase.
See what Pro includesModel prices last reviewed 2026-08-05. Providers change rates without notice, so treat every figure as a starting point and confirm against your provider's billing.