After digesting 10 hours of AI news: routing model Jev is the new hype
Muted_Ad_9442 · reddit · 2026-09-22
The author used AI to digest 16 AI news videos (10h14m) from the past month; the biggest surprise wasn't a giant model release but Jev, a fast narrow-decision model: given candidate answers upfront, it returns confidence scores — classifying customer messages or deciding inside an agent whether to call SQL, search, email, or hand off to another agent. Creator-reported figures: 70–500ms responses, one decision in 0.114s for $0.00081 vs 8.5s and $0.013 for an LLM (not independently benchmarked). Likely stack: request → Jev classifies → the right tool/LLM works → Jev checks output. The author's takeaway: the next phase may be stacks of specialized models coordinating cheaply, not one model doing everything.
Other threads: AI is merging into docs, slides, notebooks and voice assistants; reasoning models keep advancing (recently-stuck math problems now solved); the safety debate is polarizing — Nate Soares and Connor Leahy urge slowing down, Jensen Huang calls doomsday claims overblown and points to testing, engineering, product liability and existing laws.
Related event: OpenAI Researcher's Jev Goes Viral, Reviving Talk of Rescuing Old ML Ideas(7 posts)→
More from coding & agent
- Cloudflare Talk: Swapping Some Agent LLM Calls for Fast Models Like Jev Gives Big Speedups — michellechen · 2026-11-12
- Engineer Curates 15 AI Agent Design Patterns Across Single-Agent, Multi-Agent and Feedback Tiers — MaryamMiradi · 2026-11-12
- Google Antigravity rolls out browser use, closing the loop from research to coding to live verification — rseroter · 2026-11-12
- Prettier 3.10 ships new Flow/JS syntax and a Zig-powered Yuku plugin — Vjeux · 2026-11-12
- YC-backed Graphide claims CI runner 10-20x faster than GitHub on 24 repos — ycombinator · 2026-11-12
- Lessons from a 1600-page transcription: general thinking resists staying organized — fkasummer · 2026-11-12