Type in a prompt and watch, in real time, which next token (a small chunk of text) is most likely.

Use the sampling controls to influence how predictable vs. surprising the choice is:

• Temperature: a creativity dial. Lower = more deterministic; higher = more varied results.
• Top‑k: consider only the top k most likely tokens (e.g., k = 5).
• Top‑p (nucleus sampling): consider the smallest set of tokens whose cumulative probability ≥ p (e.g., 0.9), then sample from that set.

Play with the sliders to see how the distribution shifts—does it choose the obvious continuation or a plausible surprise?

When you type in any sentence or word, you’ll see:

• the list of tokens the model uses (these are often parts of words, not necessarily full words),
• for each token, its unique ID (a number) that the model uses internally.

For example, if you type “forgetting,” the model might break it into something like “for” + “getting” (these are tokens), and then show something like “1640 37210.”

Seeing this helps demystify what happens before the model “knows” anything — i.e. how text is encoded into numbers it can process.

Instead of saying “write HTML, CSS, JavaScript code,” you can describe what you want— for example, “a quiz game that asks 5 general‑knowledge questions,” or “a webpage that shows images based on user input.”

Behind the scenes, the tool translates your prompt into working code and spins up a running mini‑app.

It’s a hands‑on way to explore how prompt → code → working product workflows might function.

TravelTime Calendar helps digital nomads and frequent travelers manage their schedules across different timezones:

• View your schedule in multiple timezones simultaneously
• Convert events to local time zones with a single click
• Add, edit, and manage events with timezone awareness
• Perfect for coordinating meetings and appointments while traveling

Whether you're hopping between continents or coordinating with a distributed team, this tool takes the guesswork out of timezone conversions.