Command Palette

Search for a command to run...

Live demo
You are browsing real pipeline data captured from a live 5-Domia home mesh — latencies, delegation journeys, memories and emotions are genuine. Voices are synthetic. The demo is read-only.

Analytics

Time to first audio, per-stage latency, model performance and the labeled eval corpus.

S2S time to first audio

1.02 s

median, speech in → speech out

Interactions

57

across 2 flow types

On-device

81%

run on the origin device vs delegated

Eval corpus

10

9 good · 1 needs work

TTFA waterfall (avg, voice replies)
STT119 msLLM607 msTTS2.19 s

First audio at 1.24 s — the user hears Domia speak 1.69 s before the full 2.92 s pipeline finishes, thanks to per-sentence LLM→TTS pipelining.

TTFA distribution
Volume over time
Latency trend

Tokens & throughput

33 turns

Avg throughput

104 tok/s

Avg prompt

1162 tok

Avg completion

18 tok

Context used

28%

Throughput by model
Prompt vs completion
TTFT distribution
Skills & tools

Turns w/ tools

46%

26 turns

Tool calls

26

Error rate

0%

Turns by source
Local mic98%
Native2%
Spoken input

Avg utterance length

2.53 s

On-device vs delegated

On-device46

TTFA p50

914 ms

p95 2.14 s

Total p50

2.64 s

p95 6.21 s

Delegated (gRPC)11

TTFA p50

1.12 s

p95 1.57 s

Total p50

4.41 s

p95 11.83 s

Fastests2s405 ms

“Remember that the plumber comes on Thursday morning.”

Atlas · total 2.01 s

Slowests2s3.80 s

“Marlow, I'm so ready to unwind tonight. Recommend me something for movie night. I want a heist film.”

Marlowe · total 3.80 s

TTFA by flow
Latency breakdown
Model performance
Turns by Domia
Atlas37%21
Sous16%9
Marlowe16%9
Torque12%7
Luna12%7
Sol7%4
Eval corpus

Graded

10

Good

9

Needs work

1

Tagged

10

Export Good (JSONL)Export Needs work (JSONL)
Live node metrics

Latency percentiles and speculation counters asked of a node directly.

No Domia to query yet.