Meet Jev: The AI Built to Make Decisions

Video on YouTube by vogel

Views: 66,786
Duration: 4:07

Read from YouTube on 24 Sep 2026. Counts change daily.

Reported by the author

Avg response time: ~200 mswith average response times around 200 millisecondsyoutube.com
Speed claim tested: ~200xTypeSafe reports roughly 200× faster performance on its workflow benchmarksyoutube.com

Vogel runs Jev over their own inbox. They exported 1,500 emails and asked four questions about each: category, priority, whether it is spam, and whether it needs a reply. The first batch of 100 ran with eight parallel workers. They report an average of 200 ms per email, a p95 (the time 95% of requests finish within) of 240 ms, and about 38 emails a second. Vogel spot-checks a few results on screen: an OpenAI account notice and a domain revocation marked high priority, and a reply-needed probability of 91% on an account violation email.

A second run of 1,000 emails stays around 200 ms per email and passes 1.1 million input tokens while it runs. The usage page shows 5 cents spent. Vogel notes it is a shared team account, so that figure also covers other people’s playground calls. The latency shown is the full round trip, from sending the email to getting the answer back.

Open on YouTube

PrimitivesChoice Score Noul

Use casesHierarchical classification Support inbox triage Feature extraction for machine learning

Ran Jev against an existing classifier eval that previously used Gemini 2.5 Flash Lite

Vercel CTO reports Jev saturated an existing classifier eval that had used Gemini 2.5 Flash Lite and ran about 6x faster. Original post; TypeSafe's quote-tweet is a separate item.

Malte Ubl (@cramforce)Open

220k viewsSpeed vs Gemini 6x faster

ProjectHierarchical classification+1

Introducing Clef: our open-source decision models, and new RL fine-tuning platform

Cloudflare's launch post for Clef and Clef-flash, two decision models on Workers AI that accept the Jev request format and open weights under Apache 2.0. It covers the architecture, Cloudflare's own benchmark tables against Jev, Kev and Laya, and a new reinforcement learning service.

Michelle Chen (@michellechen)140k viewsOpen

ArticleSupport inbox triage

GLiNER: generalist lightweight NER

Compact encoder model that takes the label set as input at inference time and extracts arbitrary entity types zero-shot on CPU. Same classifier-that-takes-its-categories-as-input shape, for extraction rather than decisions. Star count is GitHub's rounded display figure.

urchade3.9k starsOpen

AlternativeStructured extraction+2

SetFit: efficient few-shot text classification

Fine-tunes Sentence Transformer embeddings plus a lightweight head for high-accuracy classification from a handful of labeled examples, with no prompting. The standard cheap alternative to an LLM classification call when the label set is fixed. Star count is GitHub's rounded display figure.

huggingface2.8k starsOpen

AlternativeSupport inbox triage+2

GLiNER2.5-Decide

Open-weights 340M-parameter decision model from Fastino Labs. It jointly decodes typed classification questions and, per Fastino's own internal benchmark, leads 9 of 17 datasets against JevK5, SemIf and Laya.

Fastino Labs (@fastinoAI)Open

2.2k starsCPU latency 167 msGPU latency 38-47 ms

ToolIntent and model routing+1

ModernBERT

Modernized BERT encoder used as the backbone for fast fine-tuned classifiers and rerankers. The non-generative baseline Jev's cost and latency claims are usually measured against. Star count is GitHub's rounded display figure.

AnswerDotAI1.7k starsOpen

AlternativeSupport inbox triage+3

Menu