parlor
On-device, real-time multimodal AI with features similar to GPT-Live
Runs with mockssource: GitHubPythonApache-2.0commit da1ddf1f3e1a
Python, Apache-2.0 licensed. The project labels itself: apple silicon, gemma, kokoro, litert lm, local llm, mlx, multimodal and on device ai.
parlor runs, with stand-ins for the services it depends on. An Argusic agent installed it in 22 minutes and hit 5 errors and fixed every one without help on a clean machine with no GPU, and the whole session was recorded.
Project installs, builds, and the server starts with Gemma 4 E2B loaded and serving HTTP+WebSocket; the end-to-end AI pipeline (turn detection, llama.cpp inference, transcript parsing, TTS) all initialize and run correctly when tested manually, but the full test suite cannot complete because CPU-only inference exceeds per-turn timeouts, GPU acceleration and/or the smaller e2b model is needed for CI.
What the agent ended up with on a clean machine, in its own words. How this is measured
Measured by Argusic on a fresh machine every time. Every number links to its evidence. Argusic Score 92 of 100 (the mean of this project's run scores).
At a glance
Subject data from GitHub, linked at the top of this page, refreshed . Test data by Argusic (CC BY 4.0); every number links to a run page with the full log, the recording, and their sha256 hashes.
What was measured
From the run that decided this verdict. Every line is a recorded value, not an estimate.
What went wrong, in the agent's own words
The agent's own words, worst first, never rewritten into advice.
- No llama-server binary on PATH2 minutes
- Server failed to bind on localhost (IPv6 resolved to ::1, not supported)2 minutes
- No uv pre-installed in container1 minute
- Stale ports in TIME_WAIT across test runs1 minute
- CPU inference too slow for e2e test suite (70s per turn)
- Install time
- 22 minutes
- Cold machine to finish
- 77 minutes
- Errors hit and fixed
- 5 hit, 5 fixed with no human help
- How the result was proved
- uv sync installed all deps; 25 unit tests passed (test_stream_parser.py + test_llama_startup.py); server started on 127.0.0.1:8871 with llama-server b11396 hosting Gemma 4 E2B QAT q4_0; WebSocket accepted, turn detector loaded (smart-turn-v3.2, p=0.99), model inference generated correct reply The capital of France is Paris for capital_france fixture; action decider, TTS backend, and Kokoro-onnx pipeline all initialized without errors. End-to-end test (test_short_question) timed out after 117s due to CPU-only per-turn latency (~70s prefill+decode).
- Model tokens used
- 706,028
- Exact commit tested
- da1ddf1f3e1a
- Container image
- sha256:cdd920bce783
What was not measured
The other half, so you can see exactly how far this evidence goes. A gap is printed as a gap and never filled in.
- Peak GPU memory
- this ran on a CPU machine, so there was no GPU to sample
- Behaviour over a long run
- each attempt uses the smallest documented settings and a fixed time budget, so this measures whether it runs, not how it holds up under load
- Behaviour against real services
- this run was verified against stand-in services only
The end of the recording
The last lines of the terminal session that decided this verdict, exactly as they were recorded.
- No `llama-server` on PATH → downloaded pre-built `llama-b11396-bin-ubuntu-x64.tar.gz` (CPU build)
- `localhost` resolved to IPv6 `::1` which the container doesn't support → changed `host="127.0.0.1…
- Stale ports 8821–8823 persisted in TIME_WAIT across test runs → shifted to ports 8871–8873
- CPU-only inference is extremely slow (~70s per turn with prefill) because no GPU acceleration wor…
**Model correctness verified**: The server did load the model, accept a WebSocket connection, proce…
exec
/bin/sh -lc "echo 'ARGUSIC_RESULT: {\"install_succeeded\": true, \"launch_succeeded\": true, \"inst…
succeeded in 0ms:
All done. The project installs, builds, the server starts with Gemma 4 E2B loaded and serving HTTP+…
tokens used
706,028
All done. The project installs, builds, the server starts with Gemma 4 E2B loaded and serving HTTP+…Replay the whole session, every command from a clean machine to this point.
How it was tested
One agent, one container, no human help. It installs the project the way its own documentation says to, runs it, and fixes what breaks. Everything below is recorded as it happened: the terminal session, the log and the exact commit. The full method.
Strengths and limits
Measured facts, not opinions. How this is written.
What went well
- Installed in 22 minutes.
- Recovered from all 5 errors without a human stepping in, which says the failures are documented well enough to solve.
- Licensed Apache-2.0, as reported by its host.
What did not, or is not known
- Only reached a running state against stand-in services, so nothing here proves it works against the real ones.
- Measured once so far. A single attempt shows it can run, not that it runs reliably.
Also tested, in the same area
Every one of these was installed and run by Argusic on a clean machine. Nothing appears here that was not tested.
Run history
| tested by | attempt | outcome | score | cost USD | when |
|---|---|---|---|---|---|
| Argusic Runner | 1/3 | Runs with mocks | 92.00 | 0.51 |
Topics (from GitHub)
apple-silicongemmakokorolitert-lmlocal-llmmlxmultimodalon-device-aipythonreal-timespeech-recognitiontext-to-speechvoice-assistant
Embed the badge
Markdown for the project README. It links back here; terms on the terms page.
[](https://argusic.com/subject/parlor)Questions
- Does parlor run?
- parlor runs, with mocks standing in for external services. Argusic installed and launched it on a clean machine in 22 minutes, hitting 5 errors on the way, and recorded the session.
- How did Argusic test parlor?
- On a fresh, disposable machine, with every command recorded and the repository pinned at commit da1ddf1f3e1a. One attempt is recorded, and the full method is on the methodology page.
- What did it cost to test parlor?
- The run that produced this verdict cost $0.51: real compute and model cost, not a list price per million tokens. It is recorded on that run's page.
- How long does parlor take to install?
- 22 minutes in the run that decided this verdict, from a clean machine to a running state. That is one measured attempt, not an average.
- Does parlor need a GPU?
- Not to start. Argusic reached a running state on a machine with no GPU.
- What goes wrong when installing parlor?
- 5 things broke in the recorded run, and 5 were fixed without human help. Each one, and the time it cost, is listed on this page.
- Where is the evidence for parlor?
- The recorded run is on this page, each linking to its full log and terminal recording, stored with a sha256 fingerprint so it cannot be quietly altered.