latency ledger
Every run, measured
runs logged
0
retrieval path · ragMsover
—ms p50
- p70
- —
- p90
- —
- p100
- —
- mean
- —
speech → text
—ms p50
- p70
- —
- p90
- —
- p100
- —
- mean
- —
guard
—ms p50
- p70
- —
- p90
- —
- p100
- —
- mean
- —
embed
—ms p50
- p70
- —
- p90
- —
- p100
- —
- mean
- —
retrieve
—ms p50
- p70
- —
- p90
- —
- p100
- —
- mean
- —
first token
—ms p50
- p70
- —
- p90
- —
- p100
- —
- mean
- —
generation
—ms p50
- p70
- —
- p90
- —
- p100
- —
- mean
- —
wall clock
—ms p50
- p70
- —
- p90
- —
- p100
- —
- mean
- —
latency benchmark
30 real dataset queries through guard → embed → retrieve, measured server-side on this instance.
0300 ms
recent queries
0No runs logged yet. Ask something on the console and it lands here.