BRAIN AUTO REAL

One request.
The cheapest intelligence available.

Anyone can provide compute. Anyone can buy it. BRAIN estimates every execution target from measured latency and configured prices, picks the cheapest valid one for your priority, executes, verifies where it can, and issues a permanent receipt.

RequestREAL

Workload

Parallel u32 matrix multiplication split across every real node online (0 now), spot-check verified by the server. Not model inference.

Priority

Routing mode override

No real nodes online. Open /node on a device and join first.

POST /v1/chat/completions
{ "model": "brain/auto",
  "priority": "cheap",
  "messages": [...] }

Route evaluationREADY

Submit a request. Every execution target is estimated from measured latency and configured prices; unknown values are shown as UNKNOWN and penalised, never guessed.

One request.The cheapest intelligence available.

One request.
The cheapest intelligence available.