Cost and rate control with decision models

CLI that scans TS/JS and Python code for LLM calls that are really decisions, converts them to Jev and shadow-tests them on real traffic.

Python library that packs many items into each Jev request for bulk yes/no classification and raises on any unanswered item.

Tiny demo that asks Jev whether an agent should continue, downgrade its model, or stop, given budget and plan.

Proxy that distils a repeated Jev classification task into a local model on the fly and answers most requests itself once it agrees.

Single-binary local cache that memoizes Jev-class decisions on model, schema and redacted state, for cheap repeats and CI replay.

Gateway that distils System One or LLM answers into small fine-tuned classifiers and serves them when confident, falling back otherwise.

Independent measurements of Jev versus open Laya on typed decisions, plus a local-first cascade that matches Jev at 1.8x speed.

Claude Code plugin that routes each prompt, step and subagent to the cheapest capable Claude model with effort caps and a live band.

CLI that fits per-question confidence thresholds to a target accuracy on your labelled data, reports escalation share and fails CI on drift.

OpenAI/Anthropic-compatible Go gateway where Jev or Laya rates topic, complexity, risk and privacy to pick the cheapest adequate LLM.

Web search CLI for agents that has Jev score results and scraped chunks against the query and goal, passing only relevant ones on.

Gate that asks Jev whether a timer or event is worth resuming a sleeping agent's LLM, skipping only when confidently not.

Reproducible 2,976-request measurement of whether batching questions in one Jev call bills the state once and keeps answers stable.

Go event bus whose optional routing layer asks Jev a Choice to pick an event projection only for ambiguous cache misses.

Halv blog post reporting a 57.1% agent cost reduction from putting Jev in the agent loop on SWE-rebench Astra pairs.

Blog post that replaces an agent's classification loop with a single Jev call and reports the loop running 7x faster.

OpenRouter cookbook: Jev checks a cheap model's draft against grounding constraints before accepting it or escalating to a stronger model.

X post by a WebMCP benchmark author reporting Jev plus a small LLM solved 100% of tasks at about 112x lower model cost.

X post working through the economics of solving CAPTCHAs with Jev at $0.0068 per hundred versus a market paying a cent each.

X post listing nine jobs where a typed Jev decision can replace an expensive full LLM call.

X account of @moritzkremb, cited in a Chinese Jev list as the source of its most specific per-call cost figure.