317 projects tested, and "No Rust toolchain installed" stopped 8: week of September 14, 2026

317 projects installed and run on clean machines in the week of September 14, 2026, 237 verified against the real thing. Every number links to a recorded run.

Covering 2026-09-14 to 2026-09-21. Published .

What was measured

Projects run
317 over 335 attempts (226 the week before)
Verified against the real thing
237
Verified with stand-ins only
67
Could not be verified
31
Tested here for the first time
10
What it cost us
31.61 USD, 53% of it model tokens

What it ran on

Argusic publishes other people's server prices every day. These are its own: the machines this week's work was rented on, at the rates actually paid, each one destroyed when its job finished.

Attempts on CPU machines
336 at 0.1200 USD an hour
Data centers used
EU-CZ-1, EU-NL-1, EU-RO-1, EUR-IS-1, US-CA-2, US-GA-2, US-KS-2, US-MD-1, US-MO-1, US-MO-2, US-NC-1, US-NE-1

What the week cost, broken down

A single total answers nothing for somebody working out whether they could do this themselves. The difference between the dearest and the cheapest project is how much fixing each needed, not a price anybody chose.

Model tokens
16.61 USD
Machines
15.00 USD
Per project measured
0.10 USD
Dearest project
career-ops at 0.51 USD
Cheapest project
codex-with-chatgpt at 0.00 USD

What the pool looks like now

The state of the whole body of evidence at the end of this week, not only what the week itself added.

Projects measured in total
480
Of those, reached a working state
380
Most measured languages
TypeScript (121), Python (113), Rust (65), Go (53), JavaScript (42)

What kept breaking

Counted across projects, not across attempts, so one repository retried three times is one problem. Each label is the wording an agent actually used.

The problems that stopped the most projects in the week of 2026-09-14
8 projects
No Rust toolchain installed
4 projects
pnpm not found
3 projects
Go compiler not installed
3 projects
pnpm not found in PATH
2 projects
Go compiler not found in container

The result most worth reading

vidcutter scored 100 of 100, after hitting 12 errors that the agent fixed with no human help.

VidCutter 6.0.5.2 GUI launches under offscreen Qt platform, connects to real libmpv 0.37.0 with FFmpeg 6.1.1, shows version and help output, and logs runtime configuration from the playback engine.

The agent's own words, from its recorded report.

What this week does not tell you

Behavior over time
each attempt uses the smallest documented settings and a fixed time budget, so this measures whether a thing runs, not how it holds up under load
Anything not tested
only the 317 projects above were run this week; a project absent from this report was not measured, not judged
Real-service behavior for 67 of them
those ran against stand-in services, never the real ones

The report as written

Argusic installed and ran 317 projects on clean machines between 2026-09-14 and 2026-09-21, over 335 attempts, 91 more subjects than the week before.

237 were verified against the real thing, 67 only against stand-in services, and 31 could not be verified at all.

10 projects were tested here for the first time.

The same problems keep appearing across unrelated projects. The most common this week was "No Rust toolchain installed", which stopped 8 projects. Also recurring: "pnpm not found" (4), "Go compiler not installed" (3), "pnpm not found in PATH" (3), "Go compiler not found in container" (2).

The result most worth reading was vidcutter, which scored 100 of 100 after hitting 12 errors that the agent fixed with no human help. In the agent's own words: VidCutter 6.0.5.2 GUI launches under offscreen Qt platform, connects to real libmpv 0.37.0 with FFmpeg 6.1.1, shows version and help output, and logs runtime configuration from the playback engine.

Running all of this cost 31.61 USD, of which 53% was model tokens rather than machines.

16.61 USD of that was model tokens and 15.00 USD was machines, which is 0.10 USD for each project measured.

The dearest project of the week was career-ops at 0.51 USD, the cheapest codex-with-chatgpt at 0.00 USD. The difference is how much fixing each one needed, not a price we chose.

The work ran on machines rented by the hour and destroyed afterwards: 336 attempts on CPU machines at 0.1200 USD an hour. Those are the rates Argusic actually paid, not list prices. The pods came up in 12 data centers: EU-CZ-1, EU-NL-1, EU-RO-1, EUR-IS-1, US-CA-2, US-GA-2, US-KS-2, US-MD-1, US-MO-1, US-MO-2, US-NC-1, US-NE-1.

Across everything measured so far, 480 projects have been installed and run here, and 380 of them reached a working state. Most of them are written in TypeScript (121), Python (113), Rust (65), Go (53), JavaScript (42).

Every number above comes from a recorded run, and each one has its log and its terminal recording on this site.

All reports · How this is measured