A training-free layer that turns an open LLM into a decision model. It reads typed choice, yes/no and score answers from one prefill, removes option-order bias with no labels and, from a few hundred labels, calibrates or fits a closed-form head. Weights stay untouched.
An embedding model turned decision model. Qwen3-Embedding-4B with a merged LoRA, driven by the open JevEmbed toolkit, which scores Choice, Score and Noul questions from embedding similarity and returns a distribution without generating text.
Decides
choice, score, noul, classify, route
choice, score, noul, classify
Architecture
anyjev
jevembed
Fine-tuned from
qwen/qwen3-8b
qwen/qwen3-embedding-4b
License
apache-2.0
apache-2.0
Availability
Open weights
Open weights
Hosted by
—
—
Input price
—
—
Decision accuracy
77.1%
85.9%
Calibration error
0.034
—
Valid action rate
—
—
Median latency
—
—
p95 latency
—
—
Figures are from each model’s manifest; accuracy and latency are what the publishers report, on their own suites and hardware. Add a third model.
Questions
What is the difference between anyjev and jevembed?
anyjev is from Nokia and jevembed from HIT-TMG (Lychee Team). Both have open weights you can download and run. Both answer choice, score, noul and classify questions. Only anyjev answers route. jevembed is the smaller model, at 4.0B parameters to 8.0B.
Which is more accurate, anyjev or jevembed?
They report on different suites — anyjev 77.1% on LocalLLaMA/typed-decisions (2,000 held-out decisions), L2 with 300 labels per question, jevembed 85.9% on JevEmbed-Data test split (64,110 hard-label questions), final LoRA checkpoint before merging — so the numbers do not rank them. Test both on your own labelled examples.
Which is cheaper, anyjev or jevembed?
anyjev: Free (open weights). jevembed: Free (open weights). Open weights cost nothing per call beyond your own hardware.
Can I run anyjev or jevembed locally?
Yes, both: systemone pull nokia/anyjev and systemone pull hit-tmg/jevembed download the weights.
Evaluation suite
LocalLLaMA/typed-decisions (2,000 held-out decisions), L2 with 300 labels per question
JevEmbed-Data test split (64,110 hard-label questions), final LoRA checkpoint before merging
