How fast is 10 tokens per second really?
A simple HTML tool lets users visualize LLM token output speeds from 5 to 800 tokens/second
Mike Veerman built a small browser-based simulator that renders text at various token-per-second rates so developers can viscerally understand speed benchmarks. It covers the range from 5 to 800 tokens/second. Useful for calibrating expectations around model inference speed claims, but not a major industry signal.