Amanai

System status

Platform health and model catalog. Read-only checks — no chat traffic, no credits spent.

Checking systems…

Loading cached snapshot or fetching from API…

Last updated
—

Platform

Models

—

Loading model catalog…

Model cards show 1-hour success and request counts. Chat cards show 24-hour p50 upstream TTFT and measured mean output speed; other cards show 24-hour p50/p95 successful-operation latency where available. Video latency measures an API submit/poll, not the full generation job; realtime latency measures session setup, not first audio. Older embedding and rerank samples may have 1,000 ms resolution until they age out of the 24-hour window. Gray days mean there was no trustworthy health signal; they are not counted as outages.

Status legend