What are we building today?
Runs on this machine. Nothing leaves it.
Monitor
Live view of the model, its experts and the hardware
Throughput
–tok/s
decode · waiting for a request
–tok/s
prefill
Idle
waiting for a request
No activity in the last minute
Time per token–
VRAM hit–
Context of the last request–
Requests served–
Where the experts live
VRAM–
Pinned RAM–
SSD only–
What fills the GPU
Hardware
Recent requests
| Time | Status | Prompt | Reused | Prefill | Output | Decode | VRAM hit | Duration |
|---|---|---|---|---|---|---|---|---|
| No requests yet | ||||||||
Troubleshooting
A report for a GitHub issue: the versions, this PC, the engine's settings and the telling lines of its log (no API key). Read it before you share it.
MCP servers
Their tools run on this machine with your rights, when the model decides to call them (in this page's chat only; switch it off in Settings).
About
The model, this machine, and how to connect your tools
Model and engine
This machine
Connect your tools
Any OpenAI- or Anthropic-compatible client works with these addresses.
Settings
Chats, settings and the key are kept in this browser only. The light or dark look is the button beside the status in the sidebar.