Human Token Benchmarker

10-second human benchmark for typing or voice, compared with model-like token throughput.

Tokenization: o200k_base

Voice model: Vosk small en-us (local)

⚡ Tokens/s 0.00

🧱 Total Tokens 0

✍️ Characters/minute 0

Time left 10.0s

Loading tokenizer...

Input Text Voice

Loading voice model... please wait.

typing-session.tw Text input mode

Voice mode ready. First run downloads a local model (~40MB). Press Start Recording to begin a 10-second run.