Hands-on test of Typesafe.ai's new Jev model shows promise for secret detection
teyhouse · reddit · 2026-09-17
A Reddit user with early access tested Typesafe.ai's new Jev model on code secret detection, reporting promising and consistent results across multiple runs.
- Test repo and datasets built around gitleaks, GitGuardian, Decryption Digest, Cremit, plus some Claude-generated samples
- 13 categories including embeddedcredential, privatekey, password, placeholder, and redacted
- Uses the Noul primitive (boolean + probability) output
- Author suggests comparing against small language models trained for this exact task, citing Wiz's blog on SLM-based secrets detection
- Open question: whether Jev's low cost justifies adoption today
Repo: github.com/teyhouse/jev-secret-detection
More from Models
- Jev passes 8/9 computer-use tasks, makes decisions 13.6x faster than Astra/Codex — iamrobotbear · 2026-09-17
- Typesafe AI Launches Jev: A Decision Engine at 70-500ms and $0.042 Per Million Tokens — brandon_galang · 2026-09-17
- GPT-6 Astra tops Terminal-Bench 4.0 at 57.7%, Claude Opus 5 hits 51.8% at half the price — shensi · 2026-09-17
- Dev take: models are smart enough now — focus on making them fail less — rickasaurus · 2026-09-17
- OpenAI Publishes Misalignment Disclosure Framework, Plus Six Incident Reports From Six Months of Training — Thom_Wolf · 2026-09-17
- A 'Quadrillion-Parameter' Model Surfaces on X, But Details Remain Unverified — KyeGomezB · 2026-09-17