My AI Portal

GPUs

Cached Models Refresh

Load Settings

Speech-to-text (Whisper)

Whisper large-v3 runs in this web server process, independent of the chat model. First load downloads weights (several GB). Ensure ffmpeg is installed for browser WebM audio. On iPhone/Safari, voice input needs HTTPS (or localhost); plain http://192.168… will not expose the microphone — capture still happens on the user’s device, not the server.

Download Model

Server Log

No log data

System Monitor

User Management Refresh

Change Admin Password

Invite codes

Registration requires a valid code. Each code allows exactly one signup; you still approve new users as usual.

Unused codes