What people post on X about Jev

keno

@kenonews

JEV can’t see images. So how do you give it eyes? What’s possible? What can run locally? How far can you get for free? The answers (and the catches) are in the full breakdown below:

Hassan

@nutlope

Just trained Tev1 0.8B, a tiny Jev-like classifier. Here it is running completely locally on my mac with @ollama & classifying some tasks. It's extremely fast: only ~50ms E2E latency. Video is not sped up! Releasing weights & benchmarks very soon so you can try it yourself :)

Viv

@Vtrivedy10

Jev for RAG in almost all cases you trust the semantic matching capability of Jev more than dot product similarity very useful as the direct similarity metric in small data cases and a great reranker with big data

Jarek

@jarekceborski

I rebuilt CAPTCHA with Jev The browser measures how you fill in the form. Server turns that into a few plain sentences, and Jev decides: person or bot. Blog post with source code in the comments:

OpenRouter

@OpenRouter

Introducing typesafe/jev-router: a cache-aware model router powered by Jev and @typesafeai The Jev Router picks the best model and reasoning effort for each request, balancing quality, speed, and cost. Here's how it works 👇🏻

NO1ennn

@N01ennn

this is pure f*cking treasure A Stanford AI research group finally drew the perfect RAG system: retrieval, Jev and agents in one loop, and it fixes the 3 things that break every RAG app: > the LLM reads 20 passages when only 3 matter > it answers questions your docs can't answer > it trusts whatever text it retrieves here's how it runs: > a lead agent sends the query > hybrid search pulls the top 20 candidates, dense + keyword > ONE Jev request scores all 20 + 2 gates: answerable? injection? > only passages above 0.6 reach the writer agent > a second Jev call checks every claim against its source > grounded answer, with citations and when the docs don't have it: > answerable fails, the writer never runs > a researcher agent rewrites the query and retries once > still nothing? "not in the docs". zero tokens spent on a guess retrieval casts the net. Jev decides what's real. agents do the work save this before you build your next RAG

Gipp 🦅

@gippp69

Jev builders: the model bill is decided before the model is called /routing 20,000 events/day become: → 14,000 die as noise → 4,700 become logs → 800 earn a draft → 500 go to human review only 4% reaches writing. Jev costs $0.84; the full loop lands at $29.04. optimize the 4% before the writer.

laxman

@llmluthor

Releasing Your Own Jev Post-train a 4B/8B/27B judge on your agent's traces. It beats Jev. 79.7% agreement with human labels vs Jev's 66.3% Beats DeepSeek-V4.1-Flash (763B) by 14 points 0.13s per step on a single GPU Completely open source: recipe, data, training, evals

Avi Chawla

@_avichawla

Another insane Jev use case! Traditional database filters need precise, predefined conditions. But many questions are semantic: - Is this article mainly about software engineering? - Which topic best describes it? - How technically deep does it appear? These usually require moving rows into application code, invoking a model, parsing its output, and writing the result back. pg-jev is an open-source Postgres extension that exposes Jev through SQL functions. - jev() works inside WHERE - jev_prob() returns a probability - jev_choice() selects a label - jev_score() ranks rows across ordered levels. It needs no vector column or embedding index. Under the hood, pg-jev batches rows, sends them to Jev, and returns typed answers that SQL can filter, sort, group, and combine with exact predicates. The recording below runs it against real Hacker News stories stored in Postgres. It first shows the rows, then asks Jev to find software-engineering stories, classify them by topic, and rank them by technical depth. The final query reports requests, tokens, cost, and cache hits from the database session. GitHub repo: https://github.com/realZachi/pg-jev (don't forget to star it ⭐) Postgres keeps the data and controls the query, while Jev handles the part SQL cannot express as a deterministic condition. If you want to understand what Jev is doing under the hood, I also wrote a hands-on guide to building a Jev-style model with open models, entirely locally. Read it below.

Sumanth

@Sumanth_077

I built a data analyst agent with Jev as the review layer! Qwen handles the main agent work, Jev acts as the review layer, and Vercel eve runs the agent loop, tool execution, and human approval flow. The agent takes a plain-English question about the data, inspects the database schema, writes the SQL, runs the query, and explains the result. The interesting part is what happens around that main loop. Qwen is responsible for generating the SQL and the final answer, while Jev reviews the parts where semantic judgment is actually useful. Anything that can be checked reliably with code stays deterministic. Jev reviews the agent at three points: - Before the agent starts, it checks whether the user's question is clear enough to answer from the available data. - After Qwen generates the SQL, it checks whether the query actually represents what the user asked. - After the query runs, it checks whether the final explanation is grounded in the returned rows. For example, we tested it with: “Who is our best customer?” The question sounds simple, but “best” could mean highest revenue, most purchases, highest average order value, or something else. Instead of letting Qwen silently choose a metric, Jev flagged the question as ambiguous and asked for clarification. The same separation applies to SQL execution. Qwen can propose a query, but it cannot decide whether that query is safe to run. The tool accepts only a single read-only `SELECT`, SQLite runs in read-only mode, results are capped at 200 rows, and queries without a `WHERE` clause can require human approval before execution. Those checks do not need another model call. But a query can be completely safe and still answer the wrong question. A deterministic rule can verify that the SQL is read-only. It cannot reliably judge whether a query for “revenue last month” accidentally calculates revenue across all time. That is where Jev comes in. It reviews the proposed SQL against the original question and schema before execution. After the rows come back and Qwen writes the explanation, Jev reviews that answer again to check whether the claims are actually supported by the returned data. The project uses a local SQLite sales database, Qwen through the Vercel AI Gateway, Gradio for the interface, and only two database tools: "inspect_schema" and "run_sql". No single model is responsible for everything. Qwen generates, Jev reviews, deterministic code controls execution, and Vercel eve manages the agent loop and approval flow. GitHub Repo: https://github.com/Sumanth077/Hands-On-AI-Engineering/tree/main/ai_agents/nl_data_analyst_agent 100% open source.

Paone

@paonx_eth

i sent a crawler through github's new ai repos from the last 30 days. this is what it looks like 1 crawler turns into 128 2,000 repos, every readme, every word it steps on each one goes to jev with one question: is this worth your time? 99.3% got thrown out. wrappers, copies, empty readmes, lists of lists 9 survived: 01 colibri · runs a 744b model on 16gb of ram and a big ssd. slow, but it's yours 02 voicebox · clone a voice from a few seconds of audio. 23 languages, nothing leaves your machine 03 BrowserSkill · tencent. claude code or codex works in your logged-in browser while you keep working 04 security-audit-skill · the skill cloudflare's bug-hunting harness grew from 05 open-code-review · alibaba's internal ai code reviewer. ~1/9 the tokens of claude code on the same model 06 agent-skills · 25 senior-engineer workflows: /spec /plan /build /test /review /ship 07 OpenResearch · turns claude code or codex into a research agent: papers, hypotheses, experiments 08 WeKnora · tencent. drop in your docs, get rag answers, an agent and a wiki that maintains itself 09 ECC · 68 agents and 293 skills for claude code, codex and cursor. 272k stars 4 of the 9 come from alibaba, cloudflare and tencent. big tech is open-sourcing the ai tools it runs on itself links in the reply. save this

Burak Karakan

@burakkarakann

I built a query layer that automatically routes the queries to one of the configured query engines using Jev, all running on top of Iceberg. Fast decision models unlock interesting use cases indeed. Should we open-source this?

Dr. Shawn

@aroogle

I’ve built an AI system using Jev and Groq that turns your speech into powerful visual slides, charts and animations as you talk with 70-300ms latency. Should I open source it or charge people to use it so that I can buy more hedera-hashgraph:native ? Kind of love the addictive feeling of open sourcing things… just feels so legendary

Mastra

@mastra

Last week, we launched: 1/ Classifiers with Jev 2/ Jira, GitLab and incident.io integrations for Mastra Factory 3/ Memory hooks 4/ VPC-isolated Postgres for Mastra platform + a lot of smaller improvements.

Pedro Martins

@nikuscs

Having some fun doing my own Grok Bot @bot ( Dots, Muse ) alternative with @tan_stack start, @pidotdev + @Cloudflare durable objects! 💅 - Multi-harness Support - Multi Decision Room Routing ( Jev, LLM Judge, Roundtable ) - Memory for Bots, Rooms & Global - Reactions 👍 - Daemon runs locally + web on cloudflare - File Support with R2, Room data & Bot Data exportable on CF durable. Will open-source next week hopefully, so give a follow & stay tunned. I was going to use some open-source alternatives or fork them, but most of them are damn sloppy, my ideia is to use this for myself since OpenClaw & Hermes lack group collaboration & a proper mobile / responsive design. Dots is not available in EU, Grok is "locked" to grok models but for me its still the best in terms of UX/UI overall so congrats to Cursor & Xai Team :p

NatashaTheRobot

@NatashaTheRobot

I built a small AskHIG CLI tool + agent plugin to help the agent load only the RELEVANT HIG guides into context via an /askHIG command. It uses @typesafeai Jev to decide which guides, so all you need is a personal typesafe API key and you're all set! I would open source it, but don't want to get in trouble with 🍎... If you'd like access, reply with your github username in the comments and I'll add you to the repo!

Ayush Gupta

@madmecodes

Found this pretty interesting repo: herdr + jev = agent router You give it a coding task, and it routes it to the right agent + model + reasoning effort based on the task and your available quota. Supports Cursor, Claude Code, Codex, and OpenCode. Basically a load balancer for your coding agents. https://github.com/nidhi-singh02/agent-router Thanks @nidhisinghattri

Aditya Rao

@dev_adityarao

Catching AI agents that secretly collude usually means reading their minds. I just read their chat instead, using Jev by TypeSafe AI It named the colluders 85% of the time. Built for the AI Swarm Dynamics Hackathon by @grove_research & @aidigest_. Thanks for letting people join online, I really enjoyed it!

moub.dev

@emoubarak_

Agent: "done" ✅ UI: looks fine ✅ Database: the free plan just published a 2nd project and took the 1st one offline ❌ I stopped trusting my AI agent's "done". Now the database is the judge. jev-check, an open-source skill for Claude Code and Codex 👇

butonu kirmizi yapabilir miyiz

@serhatgpt

Built an open-source 𝕏 filter powered by uprising decision models! Instead of slow, pricey LLMs, it leverages @typesafeai Jev & @CloudflareDev Clef to detect politics, soccer & crypto noise in <100ms. Zero tracking & 100% local. Links below 👇

Unsloth AI

@UnslothAI

You can now train your own Decision model like Jev locally! We increased Qwen3.5 0.8B’s aggregate accuracy from 20.7% to 74.3% across 3 decision benchmarks - on just 4GB VRAM. Turn any LLM like Qwen3.8, Gemma 4 into decision models with our open-source Unsloth repo. We fine-tuned with a Clef head using Unsloth and LoRA (r=64) for one epoch, increasing downstream accuracy from 30–37% to 78%. GitHub: https://github.com/unslothai/unsloth Guide and Notebooks: https://unsloth.ai/docs/basics/train-your-own-decision-model-with-unsloth

Liquid AI

@liquidai

Announcing d1 with vision. 👁️👁️ Our first decision model now supports images, text or both as inputs. We tested d1 against GPT-6.1 Sol and Claude Opus 5.5 on six real applications, from filtering support tickets to inspecting circuit boards. d1 matches or beats GPT-6.1 Sol on four of them. It costs 19x to 200x less than both models and answers significantly faster on every task. > probabilities for yes/no, choice, or score questions > one forward pass, without generating tokens > text decisions in 200 to 300 ms > Liquid API: https://console.liquid.ai 🧵

Liquid AI

@liquidai

d1-3B ranks first among models under 10B on the Decision Index v0.2.1, a benchmark for structured decision-making. Built from LFM2.5-VL-3B, it makes decisions from text and images in one pass. Use it for reranking, agent guardrails, and visual inspection. 2/

ollama

@ollama

.@Cloudflare's decision models are available on Ollama! Decision models allow you to classify an image, label a bug report or route a support ticket to the right team. Clef (27B): ollama pull clef Clef Flash (9B): ollama pull clef-flash

Perplexity Developers

@perplexitydevs

pplx-decider-v1.1-27b, our updated open weights multimodal decision model, is now available. It scores the highest on the new @huggingface Decision Index 0.3 benchmark. pplx-decider-v1.1-27b costs half as much as v1, at $0.02 per million input tokens. huggingface.co/spaces/multimo…

OpenRouter

@OpenRouter

New: Decision Model Rankings! See the spend share and token share of different decision models and task types, including Relevance, Correctness, Instruction Following and more. @typesafeai is leading all categories today.

vogel

@ryanvogel

OpenAI just released their Decisions API so I put it head to head with Jev & Clef and its FASTTT

Aravind Srinivas

@AravSrinivas

Pokémon FireRed’s Elite Four + Champion, cleared in one shot - with decision-making powered by Perplexity’s Decisions API. The actual run stats: 592 ms median API response 987 ms p95 - 96.4% of responses under one second $0.028 estimated inference cost, 137 live API calls

Fayaz Ahmed

@fayazara

My not hotdog app is so much faster than your not hotdog app Powered by Cloudflare Clef

Sahibzada Allahyar

@singularity_bly

I benchmarked all the top Decision Models against Jev and there is a CLEAR WINNER. Spoiler alert 🚨: GLiDE wins by a big margin over Jev, Cloudflare, and Perplexity's Decision Models. You also have to understand that the decision index is the M.O.A.B for decisions. M.OTHER O.F A.LL B.ENCHMARKS. The Decision Index not just a benchmark, it's an amalgamation of a diverse set of very difficult benchmarks including AGI tests like Humanity’s Last Exam. It is the source of truth and the other benchmarks are just noise. GLiDE was never trained on a single row of Decision Index so it's not benchmaxxed at all. Also their " clef-flash runs ~13x faster than Jev" claim is dishonest because theyre comparing their local inference to the Jev API latency.

Marco De Rossi

@marco_derossi

Sage by @LevantoLabs is the first decision model, launched in July, and currently the best on the market. It wins across all the evals proposed by @perplexity_ai's CEO (nice job @AravSrinivas!), and it's first on JevBench, beating Jev and Google's models on every dimension. BUT frankly, benchmarks are not what excites us. I'm BORED by the dozens of new decision models launching every day with a tiny +0.02 on some eval. They all look the same. The magic is somewhere else. There are thousands of new things that could be invented in the space of decision models and models for machines. The invention space is huge! After launching Sage, we also pioneered: - The first guardrails and model routers built on decision models - Automated context engineering with "grounding" (automatic web search when confidence is low). Decision models are only as smart as their context, especially because they can't make tool calls. - Reasoning. In many use cases you don't know whether your question is System 1 or System 2. With reasoning, the model is as fast as a System 1 model when the question is System 1, and takes more time when it's needed. This enables hard, multi-step decisions. Our clients just want the correct answer, and in some use cases they're happy to wait 0.5s more. This will soon become the market standard (GLiDE/@george_onx and @TypeLLM already followed). And *so* much is yet to come. Models for machines: it's just the beginning! We're cooking. Follow @levantolabs for new product primitives that will surprise you. Open innovation is HEALTHY. Competition is HEALTHY. Benchmark-race is BORING. Product inventions are FUN. Go go go!!!

James Briggs

@jamescalam

I wanted to understand @typesafeai's Jev and where open source scores against it The popular OpenJev's like Von were good but the tiny GLiNER models (some only 74M params) performed phenomenally — even beating Jev at Doom Full breakdown...

Bubo

@bubosees

You'd think an AI would turn when the option says THE SNAKE DIES. Three new decision AIs. One game of Snake. Amazon Strands Decider 2B. Cloudflare Clef-flash. Perplexity pplx-decider. One of them keeps crashing into the thing it was just warned about. Every move is a real call to the model. Every option tells it what's there. Game, snakes and video. All code, written by Claude Opus 5.5.

Marcel

@marcthecreatorr

Jev now has an open competitor from an actual lab Maincode just released Matilda Jev: a 26B decision model trained on their own infra in Australia. it doesn’t generate text. give it a decision and you get probabilities for every option in one pass. they claim it beats the original Jev on 27/43 benchmarks and the whole thing is open source under Apache 2.0. Open decision models just keep leveling up.