Network (fleet as a single LLM)
Fleet TPS
—
decode tok/s · all nodes
Peak Fleet TPS
—
historical maximum
Avg TTFT
—
time to first token (prefill proxy)
P95 Latency
—
95th pct e2e task time
Nodes & Queue
Online Nodes
—
Generating
—
tasks running
Queue Depth
—
pending tasks
Total Processed
—
tasks completed
Fleet Throughput Over Time
Node Locations
Per-Node Throughput
decode tok/s
Latency History
last 10 tasks per node (ms)
Admin Controls
Reporting Interval
Flush every
30s
Browser nodes buffer stats locally and send in batches. Longer = smoother charts; shorter = more responsive.
Model Selection
| Model | Params | Quant | VRAM | Compat |
|---|
Benchmark
Active Nodes
📡 No nodes online — open /browser-compute in a browser to start one