Token Lab Playground

Token Lab

AI Playground

Pick a model, set three temperatures, then write your prompt in the section below. After you run, outputs appear first; footprint metrics and charts follow so you can review usage and compare lanes.Choose a model, set temperatures, run your prompt—then scroll to outputs, footprint, and charts.

Model

Sampling

Temperature lanes

Each lane sends the same prompt at a different temperature when you run a comparison.

Lane 1

0.00

Tight, deterministic.

Lane 2

0.70

Balanced variation.

Lane 3

1.00

More creative / variable.

0 = deterministic · ~0.7 = balanced · 1+ = creative · 2 = max randomness

Compose

Prompt & run

Start here: edit your prompt, inspect tokenizer chips, and run. Results and analytics show up in the sections below.Edit your prompt, inspect tokens, run—results and charts appear below.

Last run

Wall time, token sum, and pricing for the most recent comparison (model is set above).Stats from your last comparison run.

Parallel wall time

—

Total tokens (sum)

—

Pricing snapshot

In $0.4/M · Out $1.6/M

Results

Outputs

One card per lane while generating or after a run. The page scrolls here when you click Compare.

No outputs yet

Run a comparison to stream three answers side by side—one temperature per lane.

Metrics

Prompt footprint

All zeros until a run finishes; then shows your prompt size and combined usage across the three lanes.

Input Tokens

Character Count

Estimated Cost

$0.0000

Context Window Usage

0.00%

Analytics

Charts

Per-lane token split, estimated cost, and latency from your last run—scroll up to adjust the prompt and run again.

Bars populate after your first successful run. Token and cost reflect API usage per lane.

Token Breakdown

Cost Comparison

Total request cost: $0.0000

Latency Comparison

Tip: Lower temperatures are more deterministic; higher temperatures are more creative.