Back to Compute Mesh
All demos
Live spot market

Spot prices. Live mesh.
One real job, end-to-end.

This page hits the live /api/compute-mesh endpoints. Read current per-class prices off the supply/demand curve, then run a register → post → claim → complete → settle flow that lands a Settlement-Grid receipt for the worker.

Workers
…
Queued jobs
…
Active leases
…
Settled
…
Spot prices (per 1k output tokens)
  • Phone NPU
    utilization …floor …ceiling …… / 1k
  • Laptop NPU
    utilization …floor …ceiling …… / 1k
  • Laptop GPU
    utilization …floor …ceiling …… / 1k
  • Desktop GPU
    utilization …floor …ceiling …… / 1k
  • Edge NPU
    utilization …floor …ceiling …… / 1k

Run the full pipeline

registers a fresh @demo-worker, posts a 200-token job from @demo-client, claims, completes, attests, settles.

Every call above hits the live mesh. Phase 1 inference adapters are stubbed (the worker doesn't actually run a model — it returns a deterministic outputHash); Phase 2 plugs real on-device inference and TEE attestation in without touching this page.

← Back to overviewSettlement Grid