GPU Util % utilisation
GPU Temp °C die
Unified GB of 128 · 8 GB guard
Throughput tok / second
TTFT ms · first token
throughput & first-token from the active lane
Active Lane idle no warm brain
Settings cloud lanes · eval guardrails · live — no restart

Cloud lanes (OpenRouter key)

Your Advisor runs locally — no cloud key needed. Wire your own OpenRouter key only if you also want to chat with or compare against frontier cloud models. The cloud catalog and the spend tile stay hidden until a key exists. Your key never leaves this box — Orionfold never sees it.

Cloud-eval guardrails

The bounds on a metered cloud eval lane (a non-loopback model endpoint) — the fix for an OpenRouter eval that once hung ~2.5 h holding the lane and accruing uncapped spend. A local Spark lane runs unguarded and is unaffected. Edits write a private config file the guardrail reads at dispatch time, so the next cloud eval picks them up live.