Back to Tools

Token Throughput Visualizer

Offline comparison of token rate, time to first token, and streaming cadence.

Comparison

Compare two streaming configurations under the same output length and sample text.

Shared Scenario

These inputs drive both lanes.

Sample preset Chat sample
Loading a preset also resets lane density defaults.

Lane Settings

Each lane gets its own rate, time to first token, density, and burst pattern.

Lane A

40 tok/s, TTFT 0.8 s, natural cadence

A
Burstiness
Higher burstiness keeps the same average rate but changes how the stream feels.

Lane B

20 tok/s, TTFT 0.8 s, natural cadence

B
Burstiness
Use this to compare identical averages with different delivery cadence.
Winner Lane A
Finish Delta 7.5 s
Lane A Total 8.3 s
Lane B Total 15.8 s
Lane A Display 152 chars/s
Lane B Display 76 chars/s

Preview

Ready to compare the configured lanes.

Idle

Preview animates the full configured output for both lanes.

Lane A

40 tok/s, TTFT 0.8 s, 3.8 chars/tok, natural

Ready
Elapsed: 0.0 s Visible: 0 / 300 tok Approx chars: 0 / 1140 chars 
Press Start to simulate a 300 token response.

Lane B

20 tok/s, TTFT 0.8 s, 3.8 chars/tok, natural

Ready
Elapsed: 0.0 s Visible: 0 / 300 tok Approx chars: 0 / 1140 chars 
Press Start to simulate a 300 token response.

Report Summary

Copyable text for sharing the full scenario and race outcome.