After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI? I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev • 20-200x faster • 40-400x Show more
jlink
A record-linkage library and CLI where the match rule is written in plain English: Jev answers pair questions with a probability, with local candidate blocking and match resolution.
- Category
- Repos & SDKs
- Published by
- Community
- Author
- keltokhy
- Added
- 2026-09-22
Tagscommunitypythondatacliopenrouter
Highlights
- Every candidate pair gets a probability, so you can set a threshold, require a margin over the runner-up, or carry the uncertainty into estimation.
- In its benchmarks 146,119 candidate pairs cost $2.49 in total at about 250 pairs a second, with each judgment taking roughly a fifth of a second.
- It draws a stratified sample for hand-labelling and turns the labels into precision, recall, and a calibration table with confidence intervals.
- Works from Python, the command line, Stata, and R on .csv, .dta, and .parquet, and every probability is stored so reruns cost nothing.
- Only the fields listed in on leave your machine; blocking runs locally, and --api laya or diffusiongemma keeps everything on a local server.
Quickstart
python
import jlink
result = jlink.link(
compustat, patents,
entity="firm",
on=[("conm", "assignee"), "state"],
definition="A parent company and its subsidiary are different firms. "
"A firm that changed its legal form (Inc. to LLC) is the same firm.",
left_id="gvkey", right_id="assignee_id",
)
print(result.report())Watch out
MIT-licensed with benchmark datasets keeping their own terms; needs uv and a TypeSafe or OpenRouter key, sends the listed fields for each candidate pair to that API, and restricted-use data needs local servers and documented approval.
More like this
13GitHub stars
A Python library that adds .jev accessors to pandas and Polars: ask a natural-language question per row and get labels, scores, and full probability distributions back.
Repos & SDKs#community#python#data
Communityjevframe
12GitHub stars
A Python library that reranks search results and filters retrieved documents with Jev, scoring each candidate's usefulness as evidence and dropping those below a configurable threshold before RAG context is assembled.
Repos & SDKs#community#python#reranking
Communityjev-reranker
From the community
Posts from builders shipping with Jev right now.
The launch post
Trading bot, one decision per block
I built a trading bot with Jev! Jev decides if it should "buy" or "sell", given the price feed of an asset pair, and executes real trades. It uses Monad to place the orders on Kuru's on-chain order book in every 300ms block. Demo link → jev-trader.vercel.app
Classifying 1,500 real emails
this model is actually insane at email classification i tested it on 1500 of my own emails to see how well it works and I am blown away
@CompleteSkeptic
After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI? I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev • 20-200x faster • 40-400x
Fast browser use with Stagehand
we built blazing fast computer/browser use with Jev + @Stagehanddev. this task cost $0.001 and executed at near instant speed (in a remote browser btw) the loop: observe the page, send a11y tree as state + actions as questions, Jev decides the next action, then Stagehand Show more
@CompleteSkeptic
After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI? I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev • 20-200x faster • 40-400x
LLM-as-a-judge, sped up
Jev has spoken. It picked which model is AGI. 20–200x faster. 40–400x cheaper. This could make things like LLM-as-a-judge insanely fast and nearly free. (I tried a bunch of prompts and still didn’t burn through $0.10.)
@CompleteSkeptic
After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI? I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev • 20-200x faster • 40-400x
Instant compaction with Jev
found the perfect use case for @typesafeai Jev: instant compaction in 2026, why is compaction still a summarization prompt? Jev can make it instant by scoring every tool call and dropping what’s irrelevant