Article
CommunityJev Models & Reimplementations
Zuse: A Decision Model for European Languages
Zuse is a family of three decision models (4B, 9B, 27B) that train a low-rank adapter and pointer head on frozen open-weight bases for 27 European languages. The 27B model reaches 87.5% accuracy with pooled expected calibration error of 0.016 on an 11-source European suite.
- Format
- Article
- Reading time
- 2 min
- Published by
- Community
- Added
- Nov 2026
Highlights
- Three model sizes: 4B, 9B, and 27B, trained on frozen open-weight bases.
- Targets 27 European languages across 9 task groups and 11 evaluation sources.
- 27B model reaches 87.5% accuracy with pooled expected calibration error of 0.016.
- Miscalibration is structured by task: knowledge tasks overconfident, acceptability judgements underconfident.
- On English public suites Zuse trails Jev and Kev, including Kev models on the same bases.
Questions about this entry
Short answers drawn from the linked project or page.
Zuse is a family of three decision models (4B, 9B, 27B) that read a document and return a probability for each of a fixed set of options instead of generating text. They train a low-rank adapter and a pointer head on frozen open-weight bases, targeting 27 European languages.
On an 11-source European suite, the 27B model reaches a pooled expected calibration error of 0.016, and 0.035 when weighted per source. Miscalibration is structured by task rather than language, with knowledge tasks overconfident and acceptability judgements underconfident at all three sizes.
Zuse targets 27 European languages. Reach beyond training languages is set by the base model: across 210 languages, changing the training run moved scores by at most about 2 points within a size, while languages outside the base's documented list trail by 23 to 28 points at every size.
On English public suites Zuse is behind Jev and Kev, including Kev models on the same bases. The paper reports this as a limitation of the current design.
A public demo is available at bineric.com/zuse, where the models can be tried directly.
The paper reports two pitfalls: label-word collision between topic schemes, and an injection augmentation that taught a shortcut.
The share of decisions that can be automated at a 5% error budget follows chance-corrected skill (Spearman 0.97 over 33 source and model pairs) more closely than raw accuracy (0.91).
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
X: Standard One SH (StandardThinking)
LinkedIn: What multimodal decision models actually change, and what they do not.
LinkedInJevDirectory.org
Hacker News: Decision Index 0.2.1 scores open reproductions of Jev on a frozen suite.
jackbravo4 pointsDecision Index 0.2.1 – 70 open reproductions of TypeSafe Jev's Decision ModelRead the thread on Hacker NewsHacker News: Pointer to a DiffusionGemma plus vLLM tutorial for a Jev-shaped classifier.
1 pointHow to build a Jev-style classifier with DiffusionGemma and vLLMRead the thread on Hacker NewsMore Jev resources like this
Open-weight vision-language judge that takes a task instruction plus 10 frames of a robot manipulation attempt and predicts success or failure with calibrated probabilities in a single forward pass. Released with inference code and a 20,628-prediction evaluation set.
Jev Models & ReimplementationsArticle#arxiv#open-models#evaluation
Community
Apache-2.0 Hugging Face repository for RSI-JEV-27b, a 27B multimodal decision model based on Qwen3.8-27B that answers supplied multiple-choice questions over images, returning a selected option and probabilities for all options in one forward pass.
Jev Models & Reimplementations#open-models#benchmark#calibration
Community
arXiv paper introducing Iris, a System One design that accepts image, text, or mixed candidates while leaving Jev's decision module untouched, with order-invariance, null-probability, and candidate-reuse results on a Multimodal-Mind2Web benchmark.
Jev Models & ReimplementationsArticle#arxiv#open-models#calibration
Community
A family of 0.8B-9B ophthalmic decision models that score candidate answers without autoregressive decoding, trained on RetiJev-Data: 78,158 clinical-decision and diagnosis questions and 50,074 guideline-grading questions. RetiJev-9B answers a case in 158 ms on one GPU.
Jev Models & ReimplementationsArticle#arxiv#open-models#benchmark
Community
arXiv paper testing representation invariance and abstention on measured optical spectra of silicon metasurfaces and gold nanoparticles, evaluating d3-nano, Jev 1.13, three GPT-6 Luna configurations, and Qwen3.5-2B.
Jev Benchmarks & EvaluationsArticle#arxiv#evaluation#calibration
Community
arXiv paper proposing TimesJev, a time series System 1 model in Jev's instant-decision shape. It formalizes System 1 sufficiency and learns a compressed representation preserving key frequency patterns, reporting over 62% better results and 62.9% less inference time than fine-tuned reasoning models.
Jev Models & ReimplementationsArticle#arxiv#research#benchmark
Community
From the Jev guides
Original write-ups on what this entry is for.
From the Jev AI model community
Posts from builders shipping with Jev right now.