exo
Run frontier AI locally.
Runssource: GitHubPythonApache-2.0commit 21a54c5ea023
Python, Apache-2.0 licensed.
exo runs. An Argusic agent installed it in 20 minutes and hit 4 errors and fixed every one without help on a clean machine with no GPU, and the whole session was recorded.
exo installed on Ubuntu 24.04 x86_64 with uv, Rust nightly, and Python 3.13; all non-MLX tests pass (232 passed, 5 skipped), the server starts and serves its OpenAI-compatible API on port 52415 returning HTTP 200 with model list
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 96 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.
- Rust toolchain not installed5 minutes
- uv not installed1 minute
- exo_tools module not found in test conftest1 minute
- Rust test had stale method signature (missing discovery_service_port arg)1 minute
- Install time
- 20 minutes
- Cold machine to finish
- 26 minutes
- Errors hit and fixed
- 4 hit, 4 fixed with no human help
- How the result was proved
- uv run pytest --ignore=src/exo/worker --ignore=src/exo/download exited 0, 232 passed 5 skipped; curl -s -o /dev/null -w '%{http_code}' http://localhost:52415/v1/models returned 200; server printed 'Dashboard & API Ready' banner
- Model tokens used
- 115,865
- Exact commit tested
- 21a54c5ea023
- 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
Watch the session
The whole run, from a clean machine, at the speed it actually happened. Nothing is cut and nothing is sped up.
A recording of attempt 1, run . The verdict above comes from a different attempt, so read this as one session and not as the current result. The terminal recording it was made from is the evidence, stored with a sha256 fingerprint; the video is a rendering of it. Nothing loads from the video host until you press play.
The end of the recording
The last lines of the terminal session that decided this verdict, exactly as they were recorded.
- Installed Rust nightly (1.101.0-nightly) via rustup - Built the Svelte dashboard (`cd dashboard && npm install && npm run build`) - Synced Python deps with `uv sync --extra mlx-none` (no GPU/MLX needed for core tests) - Installed `exo-tools` package that test conftest.py depends on (`uv pip install -e tools/`) - Fixed a Rust test that had a stale method signature (missing `discovery_service_port` arg) **Verification performed:** - **Tests**: `uv run pytest --ignore=src/exo/worker --ignore=src/exo/download` — **232 passed, 5 sk… - **Launch**: `uv run exo --offline --no-downloads` started, elected itself Master, printed "Dashbo… - **API**: `curl http://localhost:52415/v1/models` returned **HTTP 200** with a full model list (10… - **Chat endpoint**: returned 404 "No instance found" — expected, since no MLX backend was loaded - **Linting**: `ruff check` — all checks passed The test suite for non-MLX-dependent code is clean and the server launches + serves its API correct…
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
- Reached a running state on a clean machine, with the session recorded.
- Recovered from all 4 errors without a human stepping in, which says the failures are documented well enough to solve.
- Ran without a GPU, so it does not need one to start.
- Licensed Apache-2.0, as reported by its host.
- Measured 2 times, so the result is not a one-off.
What did not, or is not known
- Took 20 minutes to install, slower than the median of the 46 comparable projects Argusic has measured.
Of the 47 Python projects Argusic has installed and timed, exo was the 47th fastest to reach a running state, and 32 of 47 reached one at all.
What it is a reasonable choice for
- Trying it on a laptop or a small server: it reached a running state without a GPU.
Run history
Embed the badge
Markdown for the project README. It links back here; terms on the terms page.
[](https://argusic.com/subject/exo)Questions
- Does exo run?
- Yes. exo runs. Argusic installed and launched it on a clean machine in 20 minutes, hitting 4 errors on the way, and recorded the session.
- How did Argusic test exo?
- On a fresh, disposable machine, with every command recorded and the repository pinned at commit 21a54c5ea023. 2 attempts are recorded, and the full method is on the methodology page.
- What did it cost to test exo?
- The run that produced this verdict cost $0.14: real compute and model cost, not a list price per million tokens. It is recorded on that run's page.
- How long does exo take to install?
- 20 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 exo need a GPU?
- Not to start. Argusic reached a running state on a machine with no GPU.
- What goes wrong when installing exo?
- 4 things broke in the recorded run, and 4 were fixed without human help. Each one, and the time it cost, is listed on this page.
- How does exo compare with the alternatives?
- Of the 47 Python projects Argusic has installed and timed, exo was the 47th fastest to reach a running state, and 32 of 47 reached one at all.
- Where is the evidence for exo?
- All 2 recorded runs are on this page, each linking to its full log and terminal recording, stored with a sha256 fingerprint so it cannot be quietly altered.