Measured performance
What the engines actually do
Every number on this page was produced by timing a compiled binary during its build. Nothing is estimated, extrapolated or typed in by hand — the build writes the measurement to a file and this page reads that file back.
What these numbers are
- Build-time benchmarks. Each figure is a mean over a fixed number of iterations of one named operation, taken on the machine that compiled that binary.
- Not live traffic. Nothing here is sampled from a running system, and no figure on this page updates on its own.
- Means, not percentiles. The figure in each row is a mean over the whole run. Engines benchmarked with the shared
measurecommand also record p50, p95, p99 and max for that run; those live inprebuilt/<slug>/metrics.jsonand are not summarised here. - Not a league table. Each row times a different operation — a deal, a market tick, a support intake — so the unit measured is printed beside every figure. Comparing two rows to each other compares two different pieces of work.
- Not your hardware. A different machine gives a different number, so the host that produced each row is printed beside it.
- No customer data. Operator telemetry is opt-in, stays private to the vendor, and is not the source of anything shown here.
Engines measured
8 of 10
directories under prebuilt/ carrying a metrics.json
Quickest unit on record
0.23 µs
Trading Agent · one tick: book update + z-score + edge + risk check · mean of 400,000 iterations
At that rate
4.8M/sec
on win32-x64, 2026-10-11
| Engine | Mean per unit | Rate | Unit measured | Iterations | Workers | Host | Measured |
|---|---|---|---|---|---|---|---|
| Trading Agent sold as Cross-Venue Trade Flow Engine | 0.23 µs | 4.8M/sec | one tick: book update + z-score + edge + risk check | 400,000 | not recorded | win32-x64 13th Gen Intel(R) Core(TM) i5-13420H · 12 threads | 2026-10-11 |
| Real Estate Analyzer sold as Lot Line Analyzer | 2.99 µs | 334k/sec | one deal (amortise + evaluate, parse excluded) | 5,000 | not recorded | win32-x64 13th Gen Intel(R) Core(TM) i5-13420H · 12 threads | 2026-10-11 |
| Lead Gen & Scraper sold as VenueScan Lead Runner | 3.68 µs | 271k/sec | one empty scheduler task (bot bench) — not a domain measurement | 50,000 | 12 | win32-x64 13th Gen Intel(R) Core(TM) i5-13420H · 12 threads | 2026-10-11 |
| Customer Support Agent | 3.84 µs | 260k/sec | one empty scheduler task (bot bench) — not a domain measurement | 50,000 | 12 | win32-x64 13th Gen Intel(R) Core(TM) i5-13420H · 12 threads | 2026-10-11 |
| SaaS Automation sold as SaaS Automation Engine | 4.07 µs | 246k/sec | one empty scheduler task (bot bench) — not a domain measurement | 50,000 | 12 | win32-x64 13th Gen Intel(R) Core(TM) i5-13420H · 12 threads | 2026-10-11 |
| Stock & Crypto Scanner sold as Stock & Crypto Scanner | 54 µs | 18k/sec | one full scan (22 indicators over 240 bars, file read excluded) | 5,000 | not recorded | win32-x64 13th Gen Intel(R) Core(TM) i5-13420H · 12 threads | 2026-10-11 |
| File RAG Agent | 1.40 ms | 713/sec | one hybrid query (vector + BM25, top 5) over 6000 chunks | 1,000 | not recorded | win32-x64 13th Gen Intel(R) Core(TM) i5-13420H · 12 threads | 2026-10-11 |
| Game Optimizer | 2.54 ms | 394/sec | one scheduled server frame (20,000 entities, 32 players) | 300 | not recorded | win32-x64 13th Gen Intel(R) Core(TM) i5-13420H · 12 threads | 2026-10-11 |
Not yet measured
These engines ship a binary but carry no recorded measurement, so this page makes no performance claim about them. Listing them is the point: an absent benchmark is stated rather than filled in with a number from somewhere else.
webhook-engineworkflow-automation
How the architecture behind these numbers is put together: System Architecture & Pipeline Blueprint.