Loading tool…
What is tokens per second?
TPS divides generated tokens by elapsed time, capturing how fast a model streams output. It complements time-to-first-token for perceived responsiveness.
How to use this tool
Enter durations and token counts from your logs. Compare scenarios. Document baselines for regressions.
Use cases
Benchmarking providers, tuning batch sizes, explaining streaming UX to stakeholders.
Privacy and security
Editor and conversion workflows run in your browser without uploading your content to our servers for those steps. Clear sensitive data when you finish, especially on shared machines.