CommandAGI · commandagi.com/for/realtime-voice

CommandAGI
Abrir el banco de trabajo
Menú
Abrir el banco de trabajoIniciar sesión

Realtime voice

Talk to it. It talks back.

A thread's agent can be a realtime voice model. What it heard and what it said are lines in the thread, and the session's audio seconds and cost are its last record.

Open the threads appSee phone calls

One session

One engine, any realtime model.

The harness declares the model. The host opens the provider's realtime socket with your own key, and the session engine turns its events into the thread's records.

How a realtime voice session runsThe page sends the microphone's audio to this computer's host. The host's session engine streams it to the provider's realtime model, OpenAI or xAI, with the person's own key. What the model heard and what it said become notes in the thread's chat stream. The session's usage, audio seconds in and out and its cost, is recorded when it ends.Your microphonethe page, the desktop orthe phoneThe host's sessionenginepackages/runtime/local-h-ost/realtime.mjsA realtime modelgpt-realtime-2 ·grok-voice-3.0chat streamheard · saidevent: usageaudio seconds · costHow a realtime voice session runsThe page sends the microphone's audio to this computer's host. The host's session engine streams it to the provider's realtime model, OpenAI or xAI, with the person's own key. What the model heard and what it said become notes in the thread's chat stream. The session's usage, audio seconds in and out and its cost, is recorded when it ends.Your microphonethe page, the desktop or the phoneThe host's session enginepackages/runtime/local-host/realtime.mjsA realtime modelgpt-realtime-2 · grok-voice-3.0chat streamheard · saidevent: usageaudio seconds · cost

What it heard, what it said.

A live session with OpenAI's gpt-realtime-2 on this computer, run by the live test. The operator's name is left out.

test-results/realtime-host/2026-10-15/chat.jsonlreal records · a live session · 2026-10-15
22:23:53.580 note    person:operator: Hello.
22:23:53.812 note    person:operator: Can you hear me?
22:23:55.154 note    person:operator: What is 2 plus 2?
22:23:55.995 note    agent:openai-gpt-realtime-2@livebox: Yes, I can hear you. Two plus two is four.

Case study · two providers, one engine

The same question, two voices.

The same live test against two providers' realtime models, one day apart. Both heard the question and answered it. The cost a minute is the session's own usage record.

Talk over it. When you speak over the agent, the engine records an interrupted event and the agent stops. These two sessions did not test it, so no number for it is shown here.

OpenAI gpt-realtime-2: 6.45 s of audio in, 3.75 s out, $0.0091
4.3 s
xAI grok-voice-3.0: 6.05 s of audio in, 3.51 s out, $0.08
3.7 s
a minute on OpenAI, $1.29 a minute on xAI (xAI bills started minutes)
$0.13
Cost a minute of the two live voice sessions
Los números
Cost a minute of the two live voice sessions
ModelCost a minute, $
gpt-realtime-20.127
grok-voice-3.01.28
ProcedenciaArchivo test-results/realtime-host/ (this computer)Resultado case-studies/realtime/stats.jsonHerramienta the live realtime test, read by scripts/sites/case-studies.mjs realtimeSe ejecutó en This computer (dell-gb10-1), the providers' realtime APIsCuándo

Talk to your agent.

Pick a realtime agent in a thread and press the microphone.

Open the threads appSee phone calls