Jev, the 'System One' judgment model, sweeps agent dev circles in days
机器之心 · wechat · 2026-09-20
TypeSafe AI launched Jev on Sept 15, a 'System One Model' that turns program state or text into structured judgments with probabilities — and it went viral across X, GitHub and agent communities, with 37M views in days.
- Three modes: Noul (yes/no), Choice, and Score, each returning confidence; the name riffs on Kahneman's System 1 and the Jevons paradox — as judgment gets cheaper, usage explodes.
- Agent judge: LangChain's Jev-as-a-Judge experiment found 0.44s and $0.00035 per call with standout consistency vs GPT-5.6 Luna/Terra and Claude Sonnet 4.6.
- Ad analysis: Matthew Berman classified 724 live ads from 37 brands (8,724 judgments) in 40 seconds for 9 cents, 216ms median per ad.
- Other uses: fast-jev-compaction compresses Claude Code context; browser agents use Jev to pick actions (flight search demo: 7s, $0.004).
- Caveat: TypeSafe's claimed 193.6x speed and 444.6x cost gains come from its own benchmark team — early-stage reference numbers.
Related event: Jev, the Decision-Only AI Model, Goes Viral for Speed and Cost(33 posts)→
More from coding & agent
- Veteran game dev: "AI builds a full game in a day" posts are mostly hype — bendee983 · 2026-11-04
- Dev builds interactive Excel-to-dashboard with ChatGPT, shares the prompt — TermRemarkable665 · 2026-11-04
- Agents cut dev time from weeks to a day; bottleneck moved to specs and review — Suspicious_Orchid770 · 2026-11-04
- Orgo, a YC-backed platform giving AI agents their own cloud desktops, claims 1,000+ teams — Scobleizer · 2026-11-04
- Founder: Vibe-coded a PMF-successful product, then hired pro devs to fix the mess — sujingshen · 2026-11-04
- One prompt to audit cross-region traffic: dev cut trans-Atlantic latency by moving a service — DanielLockyer · 2026-11-04