# jev-ultralightspeed

Packs 32 items into one request for bulk classification and calibrates the confidence cut that sends the least-sure rows to a person, reporting 533 items a second at 89.2 percent agreement with human labels.

- URL: https://github.com/collapseindex/jev-ultralightspeed
- Section: Data and ops
- Type: project
- Stars: 12
- Pushed: 2026-09-22T06:45:48Z
- Language: Python
- License: NOASSERTION
- First seen: 2026-09-21
- Form: Library or SDK
- Useful for: Classification and extraction (0.91), Benchmarks and evals (0.86), SDK or client library (0.83)

Page: https://awesomejev.vercel.app/p/jev-ultralightspeed/