Nodes
Your machines, connected. Monitor availability and shared resources.
Connected machines
Resources offered to this mesh
| Node | State | Shared capacity | Connection | Utilization | Generation | Local models |
|---|
A conversation with your mesh.
Choose a model above and send a message.
Responses stream directly from your connected machines.
Deployment topology & diagnostics
Choose workloads and a hardware pool. Review recommendations or pin your own node assignments.
Loading shared capacity…
Deployments
Choose deployments to allocate together
Hardware pool
Only selected nodes are considered
| Use | Hardware | RAM reserved | CPU reserved | Current deployments |
|---|
Review allocation
Recommendations respect model profiles and concurrent resource budgets.
| Deployment | State | Endpoint | Error | Updated | Actions |
|---|
Live decode rate per model and node, over a rolling five-second window. Updated every second while this page is open. Rates cover routed requests only; a direct connection to an engine port is not counted.
Per model and node
| Model | Deployment | Kind | Node | tok/s | Inflight | RTT (ms) | Relay |
|---|
Per node
| Node | Status | tok/s (all models) | Models served | Relay |
|---|
OpenAI-compatible base URL
Live routing
| Model | Deployment | Id | Node | In flight | tok/s |
|---|
Operational history with opaque node IDs. No prompts, replies, raw logs, hostnames, addresses or file paths are recorded here. IDs let you compare a node over time; this is pseudonymous telemetry, visible only to administrators.
Across your mesh
Waiting for samples
Anonymous nodes
Latest sample in selected range · expand for sensor detailsRecent records
Newest 20 matching samples, connection changes and routed responses| Time | Kind | Node | Deployment | Message |
|---|
| Name | Prefix | Created |
|---|