Rl-Training · AI Beat

17 Sep 2026 · AI Beat Desk

Teaching a 4B Model to Outplan Postgres

Rohan Bansal trained a 4B parameter model to generate Postgres join-order hints via distillation from a frontier model followed by reinforcement learning with a custom anchored GRPO variant — yielding a 1.81× geometric mean speedup on 113 join-heavy queries for about $1,200 in total compute. The methodology shows how cheap specialist RL is getting for niche performance optimization tasks.

16 Sep 2026 · AI Beat Desk

Jev Doesn't Generate Text, and That's the Point

TypeSafe AI shipped Jev, a "System One" model that deliberately abandons text generation in favor of parallel typed probabilistic decisions. By giving up string output entirely, it sidesteps hallucination, cuts latency to under 500ms, and costs two orders of magnitude less than frontier LLMs on structured automation tasks — a deliberate specialization rather than another attempt to scale up a general model.