jevbooks

← All projects

jev-lab

llt22/jev-lab

Independent lab for Jev: benchmarks, use cases, and ecosystem audit.

Evaluation & Benchmarking89%Runner-up: Calibration & Research

What it is

A community-run research lab for TypeSafe's Jev (System One model). It curates official resources, SDKs, and real-world use cases, and provides reproducible benchmarks of Noul, Choice, and Score primitives, confidence gating, fan-out latency, and agent control.

How it uses Jev

Jev turns natural-language state into typed decisions: Noul for yes/no, Choice for selecting an option, and Score for ordered ratings. The lab benchmarks these primitives, measures fan-out latency and confidence-gated escalation, and compares Jev against a deterministic keyword baseline for support routing.

Primitives:choicescorenoul

Technique worth stealing

Gate on the API's confidence, not only the largest answer probability; treat probabilities near 0.5 as uncertainty.

Try it

cp .env.example .env; python3 -m unittest discover -s tests; python3 scripts/evaluate_support.py --run-name support-routing-v1

View on GitHub

judged by Jevjev-1.13.0

Evidence

Each line is one question put to Jev about the README. ≥ 0.60 reads as yes, ≤ 0.40 as no; in between Jev is not making a call.

  • Jev-centricyes0.93
  • Shows a System One patternyes0.79
  • Handles uncertaintyno0.10
  • Measuredno0.10
  • Runnableno0.11
  • Worth recommendingunclear0.53
  • Model replicano0.05
  • Problem scopescore on a 0–2 scale1.00
  • About Jevyes0.98

Signals by Jev jev-1.13.0, card written by DeepSeek V4.1 Flash from the README on 21 Sept 2026.