Send a prompt, pick a model, and it runs on our own hardware, not a data center. The answer comes back with its receipt: which model, how fast, which build.
Prompts and answers are published in the public feed. Do not send anything private.
It joins a queue on this site. You see your place in line.
The demo machine collects jobs from the queue. Nothing can reach it from outside.
Each prompt is checked, run on its own, and the answer is checked again before it is published.
Model, seconds per token, tokens written, build, and time, right under the answer.
One machine serves everyone, so each person gets a few prompts a day and answers have a length cap. If the machine is busy or offline, the queue keeps your prompt and runs it next.
