An independent Jev-shaped reproduction on DeBERTa-v3-large. One state and any number of choice, score and yes/no questions go in, and a calibrated distribution per question comes out of one forward pass. Public gold labels only.
A Gemma 4 12B fine-tune for typed decisions that also chats and reads images. Ships as GGUF for llama.cpp, holds a 64K context on a 16 GB GPU, and serves /v1/systemone next to /v1/chat/completions.
Decides
choice, score, noul, classify
choice, score, noul, classify, route
Architecture
open-jev
winnow
Fine-tuned from
microsoft/deberta-v3-large
google/gemma-4-12b-it
License
apache-2.0
apache-2.0
Availability
Open weights
Open weights
Hosted by
—
—
Input price
—
—
Decision accuracy
85.4%
85.7%
Calibration error
0.022
—
Valid action rate
—
—
Median latency
28 ms
143 ms
p95 latency
—
—
Evaluation suite
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 open-jev-deberta-v3-large and winnow?
open-jev-deberta-v3-large is from Kotoba Labs and winnow from EldanRing. Both have open weights you can download and run. Both answer choice, score, noul and classify questions. Only winnow answers route. winnow reads up to 64K tokens of state, against 512 tokens for open-jev-deberta-v3-large. open-jev-deberta-v3-large is the smaller model, at 434M parameters to 12B.
Which is more accurate, open-jev-deberta-v3-large or winnow?
They report on different suites — open-jev-deberta-v3-large 85.4% on Kotoba held-out test, seen question types (banking77, SST-5, BoolQ), winnow 85.7% on JevBench public subset (231 items), Q8_0 — so the numbers do not rank them. Test both on your own labelled examples.
Which is cheaper, open-jev-deberta-v3-large or winnow?
open-jev-deberta-v3-large: Free (open weights). winnow: Free (open weights). Open weights cost nothing per call beyond your own hardware.
Which is faster, open-jev-deberta-v3-large or winnow?
By their publishers’ figures, open-jev-deberta-v3-large answers in about 28 ms at the median and winnow in about 143 ms — measured on different hardware, so treat it as a rough guide.
Can I run open-jev-deberta-v3-large or winnow locally?
Yes, both: systemone pull kotoba-labs/open-jev-deberta-v3-large and systemone pull eldanring/winnow download the weights.
Kotoba held-out test, seen question types (banking77, SST-5, BoolQ)
