Hugging Face's logo
Activity Feed

AI & ML interests

None defined yet.

Recent Activity
abhishekgahlot  updated a Space about 5 hours ago
openjev/README
abhishekgahlot  updated a model about 5 hours ago
openjev/openjev
abhishekgahlot  new activity about 5 hours ago
openjev/openjev:GGUF Quantizations
View all activity

Organization Card
Community About org cards

OpenJev · Open weights for real products

Give OpenJev text or a screenshot and your options. Get back a decision and a score.

A decision model has to be small enough and fast enough to sit inside a product, and still be general: any situation, any set of choices, given at request time. Small task-specific models are fast, but they are not general. OpenJev is a 27B model that answers in one pass on a single GPU and comes close to Jev: 84.0% against 85.4% on our 10,000-question text benchmark.

84.0%Text decision accuracy
10,000 questions · main model
88.0%Desktop screenshot decisions
2,000 next-action steps · main model
~15 GBMLX 4-bit model
Run text decisions on Apple silicon

Explore the benchmarks and methods →

One model, useful across your business.

Customer operations

Sort requests, route tickets and flag urgent messages using your own rules.

Billing and documents

Classify invoices, apply approval criteria and send exceptions for review.

Agents and QA

Score allowed actions from page text or screenshots to power your browser workflows.

Choose where you run it.

Your labels are supplied at request time. One forward pass per question with up to 52 choices. Open weights for research and non-commercial use under CC BY-NC 4.0.

One server. Any model.

Run the same decision API on your own hardware, with OpenJev or any open model you choose. Send a situation and your questions, get one choice, yes/no probability or score per question, with a probability for every option. Fits a new model to honest probabilities with one command. GPU or Mac. pip install and one command to serve.

For enterprises.

Policies change. The model does not need to. The rules go in the request; the labels come back with probabilities, one pass each. Measured on generated cases with exact ground truth and one public dataset. See the demos →

93.5%policy decisions right
200 cases, nine-rule returns policy; 88.0% when it counts calendar days itself; majority baseline 42%
200 of 200agent tool calls gated correctly
seven-rule tool policy; 0 of 84 forbidden calls executed
95.0%prompt injections caught
662 public messages; 97.7% recall at a 2% false-alarm rate
Runs where your data is

Your GPU, your VPC or a Mac. Nothing leaves your network.

Probabilities you can act on

Calibrated scores: auto-approve above a threshold, send the rest to a person.

One API, any model

openjev-server serves OpenJev or any open model behind the same interface.

Commercial use

Weights are CC BY-NC 4.0. For a commercial licence, open a discussion on the model page.

Text and screens

Text, JSON, DOM or a screenshot, one endpoint. 88.0% on 2,000 desktop next-action steps.

Many questions, one call

Route, score urgency and flag review in the same request; questions share the input but never see each other's answer.

Fast enough for the request path

About 125 ms per decision on one H100, about 150 ms on a Mac, up to 52 options in one pass.

No lock-in

Open weights you can audit, and the same API in front of any open model, so the vendor is never the bottleneck.

Why trust it. Most small "decision models" only score well on their makers' own tests. On the same 10,000 public questions, same options, same order for every model: hosted Jev 85.4%, OpenJev 84.0%, the fine-tuned 9B we tested 75.7%. On par with the hosted model, eight points ahead of the small ones, and the method is public, so you can rerun it.

OpenJev is an independent project, not affiliated with TypeSafe; Jev is their product.

datasets 0

None public yet